Researches in Colour Vision and the Trichromatic Theory

Google

This is a digital copy of a book that was preserved for generations on Hbrary shelves before it was carefully scanned by Google as part of a project

to make the world's books discoverable online.

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover.

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the

publisher to a library and finally to you.

Usage guidelines

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing this resource, we liave taken steps to
prevent abuse by commercial parties, including placing technical restrictions on automated querying.
We also ask that you:

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for
personal, non-commercial purposes.

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the
use of public domain materials for these purposes and may be able to help.

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for informing people about this project and helping them find
additional materials through Google Book Search. Please do not remove it.

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner
anywhere in the world. Copyright infringement liabili^ can be quite severe.

About Google Book Search

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web

at |http : //books . google . com/|

QC
h -? '

. Al5'3/v

BY THE SAME AUTHOR

TENTH BDITION THOROUGHLY REVISED

A TREATISE ON PHOTOGRAPHY

With 134 Illustrations

Crown 8vo, 5s.

{Text-Books of Science)

LONGMANS, GREEN AND CO.

LONDON, NEW YORK, BOMBAY, AND CALCUTTA

RESEARCHES
IN COLOUR VISION

AND THE TRICHROMATIC

THEORY

BY

>< SIR WILLIAM DE W^^' ABNE Y

K.C.B., D.Sc, D.C.L., F.R:5!

WITH 4 COLOURED PLATES AND OTHER

ILLUSTRATIONS

LONGMANS, GREEN AND CO,

39 PATERNOSTER ROW, LONDON
NEW YORK, BOMBAY, AND CALCUTTA

1918

All rights reserved

PREFACE

The author has brought together in book form the
substance of a somewhat large number of communica-
tions which during the last twenty-five years he has
made to the Royal Society, on the subjects of Colour
Photometry and Colour Vision.

The publications of the Society are not always
accessible to general readers, more especially to
foreigners, and perhaps as a consequence the author
has frequent requests for copies of his collection of
papers. These requests are one reason for issuing this
work ; but another one, more cogent, was his wish to
show that the Trichromatic Theory of Colour Vision
does not yet require a funeral oration over its remains.
It is not by any means as moribund as some seem
anxious it should be considered, but is, in fact, very
much alive. Other theories of Colour Vision, physio-
logical and psychological, have been oflTered in the
press, in magazines, or in books ; but the one theory
which alone takes cognisance of the physical aspects of
the subject has had no such aid to publicity in recent
years. In 1891 the author published a small elemen-
tary work on the modes of measuring colour, and later,
in 1895, a reprint of his Tyndal lectures, delivered at

V

o

51395

vi RESEAKCHES IN COLOUR VISION

the Royal Institution, which gave the then state of
the Trichromatic Theory, was also published. To make
further advances, accurate measures of the three sensa-
tions existing in the spectrum colours were necessary.
These measures having been made, a new base was
established from which an attack on various problems
that had been left indeterminate could be delivered.
This present publication gives solutions of at least some
of these problems. The author has not criticised in it
any rival theory, but has confined himself to giving an
accoimt of his own researches in regard to the colour
sensations, colour blindness, retinal fatigue, and the
like. He trusts he has shown that all the pheno-
mena he has studied, and to which quantitative
measurement can be applied, are explained by this
Physical Theory. There are some after-effects of
light on the retina which, so far, do not lend them-
selves to exact measurement by physical means.
These have not been discussed.

A theory, to be one of perfection, must offer the
truth, the whole truth, and nothing but the truth.
The Trichromatic Theory offers the truth: but the
physiologists must add their quota to it to make it
the whole truth. There may be difficulties in welding
together the physical and physiological aspects of Colour
Vision to make a perfect theory, but it will be effected.

There is such a striking likeness in the behaviour
of the photographic plate with that of the retina
when subjected to the action of light, that it is hard

PREFACE vii

to believe that the chemical decomposition of sensitive
matter (which is admittedly the result on the former)
is not the result on the latter. Until the seat of visual
sensation is definitely located, no conclusion as to the
similarity of result can be arrived at.

This work is divided into Part 1. and Part II.
Part I. is elementary to some extent, and has been
the subject of lectures to students. This part need
not, of course, be studied by advanced workers, except
so far as is necessary by the references made to it in
Part II., where Colour Vision is the main subject.
Paragraphs in square brackets may be omitted in
reading.

The author has to thank Dr. W. Watson, F.R.S.,
for advice and criticism in some of his later work.
He wishes also to acknowledge the devotion which his
assistant, Mr. W. Bradfield, has shown in forwarding
his experimental work from its commencement.

CONTENTS

PART I

CHAP.

I. Introductory

II. The Eye

III. On Phenomena in Vision

IV. Colour Patch Apparatus

V. The Source op Light to use with the Apparatus

VI. The Apparatus to Alter the Intensity op the
Light ......

VII. Intensity op Spectrum Colours
VIII. The Measurement op Luminosity
IX. Complementary and Contrast Colours

X. Numerical Registration op Colour
XI. Colour Discs

PAGE
I

8
21
33
53

68

74

86

112

126

130

PART II

XII. Extinction op Colour and Light

XIII. Colour Fields

144
190
211
223
248

XIV. The Theory op Colour Vision .

XV. The Colour Sensations

XVI. Colour Sensations in Colour Discs

XVII. Change in Hue of Colours by the Addition op White
Light, and the Amount op Colour which would
BE Added to White without being Perceived . 255 \ — ^

ix

J

X RESEAKCHES IN COLOUR VISION

CHAP. PAGE

XVIII. Congenital Colour Blindness 267

XIX. Complete Red and Grben Colour Blindness . 276

XX. Incomplete Red and Green Colour Blindness. . 293

XXI. Colour Equations for the Detection of Colour

Blindness 309

XXII. Matching a Pure Colour by a Mixture of two

Colours, and a Mixed Colour Matched bt one

Pure Colour 321

XXIII. Measurement of Green or Red Sensation De-
ficiency BY MEANS OF CoLOUB DiSCS 329

XXIY. Some Cases of Uncommon Colour Blindness . 338

XXV. On Colour Fatigue 360

XXVI. Testing for Colour Blindness 398

Index 415

LIST OF PLATES

I. Spectrum Colours as named by Persons who were com-'

PLETELY OR NEARLY COMPLETELY ReD OR GrEEN BlIND

275
II. Spectrum Colours as named by a Person avho possessed

•05 of Red Sensation

III. Series of Water-colour Patches, containing among

THEM Confusion Colours 403

IV. Spectrum Colours as named by Persons with different

DEGREES OF COLOUR BLINDNESS 404

y. Spectrum Colours as named by a Person who possessed

•35 OF Red Sensation 406

XI

PART I

/

•.'dES" — i— .' "--

RESEARCHES IN COLOtJR VISION

CHAPTER I

INTRODUCTORY

In writing on the subject of colour, it is not intended to
enter into its very elementary aspect. It is taken for
granted that the reader is acquainted with it. It is
only proposed to give a brief recapitulation of the main
facts as generally known regarding colour, and to make
these the foundation of subsequent remarks.

Colours have no objective existence ; they are simply
sensations excited by light as a rule ; though an electric
current, or mechanical pressure of the closed eyes, may
also be able to give similar sensations. With these
two last we have nothing to do, and it may be assumed
that we are dealing with the sensations that light alone
excites on the retina.

It has to be remembered that all bodies can only
be recognised by the eye when they are either self-
luminous or illuminated. Self-luminous bodies are such
as the sun, the arc light, the oxyhydrogen light,
candle flames, and gas flames. There are other bodies
which are made self-luminous by phosphorescence and
by electric action, but these latter we may dismiss for
our purpose.

Pure Colours.

Pure colours are produced by definite vibrations of
the surrounding ether. For instance, the j)ure colours

2 RESEARCHES IN COLOUR VISION

of the bright lines in the spectrum of metallic vapours
are each produced by the stimulation of the retina by
vibrations of ether waves of different lengths. Impure
colours are produced by the stimulation of the retina
by more than one set of vibrations, but it is quite
possible for an impure colour to match a pure colour.
Thus the mixture of the colours about the D (sodium)
line in the spectrum will be indistinguishable from
the D light itself. The retina does not analyse the
mixed colour, but only recognises when the two are
similar. We shall see hereafter that even with one
set of vibrations the colour recognised by the eye is
not always the same, the colour being dependent on
the intensity of the vibration, which is equivalent to
saying, on the amplitude of the wave.

The pure colours are produced by sending a beam
of white light (such as that of the sun or of the electric
light) through a prism, which sorts out, as it were,
the different wave-lengths. The colours become visible
when they fall on a white screen {i.e. which reflects
white light unaltered). If the beam of white light be
sent through a narrow slit before falling on a prism
(the slit being parallel to the edge of the prism), and is
received on a lens, there will be a band of colours, each
colour being an image of the slit when received on a
screen (or viewed through an eye-piece) placed at the
focus of the lens.

(We have mentioned white light, and it may be
advisable to say at once that there is no fixed stan-
dard of white light at present. In Chapter V. we
shall see what light is most convenient to use for
experimental purposes.)

The colours seen on the screen will be red, orange,
yellow, green, green-blue, blue, and violet, blending into

INTRODUCTORY 3

one another ; and if sunlight be used for the white light,
dark lines, some more marked than others, will make
their appearance, three of these latter, known as A, B,
and C, being in the red, D in the yellow, E and b in the
green, F in the green-blue, and G and H at the two
ends of the violet. These lines are always in the same
positions in the solar spectrum, and may be looked on
as milestones from which measures of the wave-lengths
of intermediate colours can be determined.

Another method of producing pm:e colours is by
means of the diffraction grating, which is made by
ruling fine, equally-spaced, parallel lines (as many as
150,000 to the inch and even more) on glass or metal
on which white light coming through a distant slit
parallel to the ruling is caused to fall. Using a lens to
collect the rays, and receiving the focused image of the
slit on a white screen, it will be seen that, as well as a
colourless image of the slit in the axis of the lens, there
are several pairs of spectra on each side of it, the pair
nearest it being shortest and brightest ; the other pairs
being fainter and longer. The number of pairs is theo-
retically infinite ; but practically the first three pairs are
well visible on the screen. The central image is formed
by half the white light, and the first pair of images take
about a quarter for their formation, so that practically
the brightest image is, roughly, only an eighth as bright
as a prismatic spectrum of equal length with the same
amount of light passing through a slit of equal width.
Hence for a bright spectrum it is advantageous to use
the prismatic rather than the grating spectrum. What
is said here is an approximate statement, much of the
brightness of the spectra depending on the ratio of
the width of the lines to that of the space between
them.

4 RESEARCHES IN COLOUR VISION

No pigments can accurately represent spectrum
colours, but we give the nearest approach to them
that we know. For the red, the nearest approach is
ordinary vermilion (not scarlet vermilion), with which
is mixed a small quantity of permanent violet ; for
the orange, orange cadmium ; for the yellow, chrome
yellow ; for the green, a mixture of prussian blue and
aurelin ; for the blue-green, viridian, with a small
amount of cobalt blue ; for the blue, ultramarine ; for
the violet, permanent violet, to which a little blue is
added. With these colours a fair representation of
the spectrum may be painted, but it will lack purity
and luminosity. These last essentials bring us to state
the constants of colours. Besides purity and luminosity,
there is hue. When these three are known, the colour
is defined.

Colour Constants,

The hue of a colour is recognised regardless of its
purity and luminosity. In Chapter XVII. we shall find
that the hue of a spectrum colour varies with the amount
of white which it contains, the addition of white in
some cases giving a hue which is yellower than when it
is absent. The purity of the colour is dependent on
the amount of white mixed with it. In Chapter X.
we find that most colours of nature and of pigments
can be matched with one spectrum colour, if mixed
with white in varying proportions. It follows, then,
that nearly every colour except a spectrum colour
is an impure colour. The third colour constant is the
luminosity of the colour. In Chapter VIII. we shall find
that the luminosity of the spectrum has a decided
influence on the hue of a colour, and not only of the
hue, but of its apparent purity ; for there is a certain

\

\

INTRODUCTORY 5

reduction at which the colour (and even the light itself)
disappears. From this consideration of the constants
of colour, it shows how careful an experimenter must be
in drawing conclusions from the results of observations
he may have made.

Absorption and Obstruction.

We shall notice in succeeding chapters that we
send light through different media ; if the light passes
through readily, we call them transparent media. If
the light is scattered internally, in its passage through
it, they are translucent media. In some cases the
medium may be transparent to long wave-lengths
and translucent to the shorter wave-lengths, as when
light passes through a turbid medium such as water
charged with very fine particles. We have in this last
case to differentiate between what is true absorption and
simply obstruction. We shall find that the coefiicients
of absorption and obstruction may take the same form
mathematically, but not always so. As an example of
absorption we may take black glass, and of obstruction
the silver deposit of a photographic negative. In the
one case certain of the rays of white light are blotted
out and perform work in the interior of the glass. In
the other the white light itself is more or less arrested
according to the number of silver particles it encounters,
and the part that is prevented passing through may be
partially absorbed by the silver particles and the re-
mainder scattered throughout the glass.

It may be useful to point out what absorption of
light entajils. Suppose we examine a spectrum through
an orange glass, we at once see that a very little red
and orange and yellow are absorbed, but that in the

6 RESEARCHES IN COLOUR VISION

green the absorption is much stronger, and that the
blue is totally absorbed. Evidently the coefficient of
absorption increases as the blue is approached. If the
amount of light cut off by one glass from the different
spectrum colours is measured, the coefficient of absorption
can be found for each colour. If another orange glass
be placed in contact with the first, the amount of absorp-
tion of the different colours can be calculated by using
the coefficients. The addition of other orange glasses
to the first one will reduce the blue-green, green, and
yellow-green light passing through them much more than
the red, orange, and yellow, and the result will be that
naked white light, viewed through these superimposed
glasses, will appear reddish orange. This indicates that
the colours of different transparent media alter in pro-
portion to the colouring matter present. In some cases,
such as, with a solution of methyl violet, where the
green is cut out of the spectrum, the coefficients of
absorption for the blue are greater than for the red.
By increasing the thickness of fluid through which the
spectrum passes to the eye, the blue will disappear
when the red is still bright. Examining white light
through the thicker solution, it will appear ruddy
instead of violet. This phenomenon is sometimes called
dichroism.

Colours of Pigments.

The colour of pigments is due to absorption. When
a pigment is painted over a white ground, part of the
light which strikes the fine particles composing it passes
through them, falls on the white surface and again
strikes the particles, and is received by the Qye of the
observer. There is part of the light, however, which is
reflected from the sides of the particles and does not

INTRODUCTORY 7

traverse them, and also comes back to the eye. The

mixture of the coloured light and the white gives the

sensation of paleness to the colour. When the pigment

is put on so thickly that the white ground is completely

covered, the true colour of the pigment mixed with a

small surface reflection of white light is seen by the eye.

In using colour discs (see Chapter XI.), it is generally

desirable that the white card should be entirely and

thickly covered. It should be noted that, so far as the

colour itself is concerned, the light has to pass through

the pigment layer twice. If the pigment be spread upon

glass and the eye receives the transmitted colour, the

saturation will be much less, though the colour will be

the same as that seen by reflected light.

CHAPTER II

THE EYE

Before proceeding further, something must be said
regarding the apparatus with which we can perceive
colour and light. We join light with colour, as we
shall see later on. It is necessary to do so.

Structure of the Eye.

The structure of the eye may be roughly divided
into two parts ; somewhat in the same way we can divide
a camera into two parts — (1 ) the optical part, and (2) the
impression-receiving part. In the camera the first is
the lens and the second the plate. In the eye the first
is the optical mechanism and the second the retina.
The following figure^ gives a section of the eye, in
which the several parts are more or less distorted as to
relative sizes.

ScZ is the sclerotic coat shaded longitudinally, which
is continuous with e c, the transparent cornea (unshaded).
CA is the choroid coat, with (C P) ciliary process and (I)
the body of the iris, all shaded to show they are parts
of the same vascular movement. R is the retina or
inner wall, and PE pigment epithelium or outer wall
of the retinal cup. In front of the wavy line os (ora

* The general description of the anatomy of the eye is taken from
Foster's Text Booh of Physiology y and the author is indebted to its publishers
(Messrs. Macmillan) for permission to do so, and also for the loan of Fig. I
and Fig. 2.

8

THE EYE

serrata), the retina proper changes into the pars ciliaris
retinae, pc'R. Both the pigment epithelium and the
p c R are shown as con-
tinued over the back of
the iris, as well as over
the ciliary process, C P.
L is the lens and sp
the suspensory ligament.
V H is the vitreous
humour, and the dotted
line round one side of
the lens and through
VH represents a mem-
brane which indicates an
embryonic continuation
of the central artery
of the retina passing
through O N, the optic nerve connected with the brain.
O X is the optic axis of the lens.

Fig. 1.

Diagrammatic Eye.

It will be seen that the retina is really an outcrop
of the brain. The optical apparatus is complicated by
the fact that the various refractive indices of each part
of it vary. The following table shows the variations : —

Refractive index of the vitreous humour . . 1 '3365

Kefractive index of lens 1*4371

Radius of curvature of cornea .... 7'829 mm.

Radius of curvature of anterior surface of lens . 10

Radius of curvature of posterior surface 6
Distance from anterior surface of cornea and

anterior surface of lens 3*6

Thickness of lens 3*6

These measures allo\v a reduced or diagrammatic

10 RESEARCHES IN COLOUR VISION

eye to be calculated, and the rays of light which are
brought to a focus on the retina can be traced readily.

Fig. 2 shows the path of the rays on to the retina,
and through the lens as calculated coming from the

Fig. 2.

arrow X Y and forming the image Y X on the retina.
It will be seen that the image is inverted.

In Fig. 1 we have alluded to the sclerotic coat.
It consists principally of bundles of white fibrillated
connective tissue. Part lie longitudinally and part
horizontally, and present an interlaced appearance, thin
but tough. It is scantily supplied with blood-vessels.

The choroid coat consists principally of blood-vessels
and muscular and nervous elements embedded in con-
nective tissue. It nourishes the retina, and serves as
a muscular mechanism as well. The choroid coat is an
elastic coat, which the sclerotic coat is not.

The ciliary process (C P in the figure) is a con-
tinuation of the choroid coat, but of different structure.
Cells which are embedded in it bear pigment, especially
in dark eyes.

The pigment epithelium (P E in the figure) is com-
posed of plain cubical cells which are loaded with pigment
except in albinos.

The iris (I in Fig. 1) is a continuation of the
choroid coat, which has distinctive features of its own,

THE EYE 11

and ends abruptly at the pupil. It has round the margin
of the pupil muscular fibres gathered together in the
form of a ring (the sphincter iridis).

The coimea consists of connective tissue, which is
arranged in concentric layers of bundles, all placed
evenly in the same direction. These bundles and the
substance cementing them together are all transparent.
The front surface of the cornea is covered with an
epithelium, also transparent.

The ciliary muscles are between the sclerotic and
choroid coats, with roots near the iri6.

The lens L is a transparent body of a certain re-
fractive power, and possesses considerable elasticity ; its
shape may be altered by pressure, but it resumes its
original shape when the pressure is removed. The liquid
in the lens is of a globulin nature, approaching to vitellin
found in the yolk of eggs, but albumen is absent.

The vitreous humour (V H in Fig. 1) consists of
a jelly-like material containing principally water. It
may here be mentioned that the media of the eye are
fluorescent ; a condition which is said to be conducive
to seeing the ultra violet rays, though, for our own part,
this appears very doubtful, being more likely to give
a green or violet veil covering the retinal images.

The Retina.

We next have to consider the retina, and this we
can only do in a very general way, since it is very
complicated in structure. The optic nerve, as has
already been stated, is an outcrop of the brain. A
vertical section of the retina, which has an average
thickness of '15 mm., shows that it is made up of layers
superimposed the one on the other, and these layers are

12 RESEARCHES IN COLOUR VISION

very much the same throughout the retina, except at
one part, the macula lutea (or yellow spot, as it is
called), which contains a depression called the fovea
centralis.

The layer next the vitreous humour is what is called
the layer of optic fibres. The next is a layer of large
branched nerve cells. This is succeeded by a peculiar
layer, which has a close resemblance to the matter of
a part of the cerebellum. Next are two layers of what
are called closely packed nuclei. Outside of these
comes the remarkable layer of rods and cones, which
is probably the seat of visual impression, and which
is seemingly in actual connection with the optic nerve,
and this in its turn is succeeded by the layers of pig-
mented epithelium, to which we have already alluded.
Each rod consists of two distinct parts of a wholly
different nature, called the outer and lower limb. The
outer limb is a cylinder about xS^ mm. in length and
Y^^ mm. in diameter. It is transparent and doubly
refractive, and is probably made up of a very large
number of discs, of about rvhny "^^- ^^ thickness
cemented together. The cylinder is sensitive to light,
swelling up under its action, and shrinking again when
the light is removed. It is coloured with a pink matter,
which is called visual purple, and which bleaches under
prolonged exposure to light. The inner limb is trun-
cated, and tapers to a delicate thread, which eventually
connects with the optic nerve.

The cone, like the rod, consists of an inner and outer
limb. The outer limb is conical and not cylindrical in
form, and is about y^ mm. in length. The inner limb
is very like the inner limb of the rod. Excluding the
macula lutea, the rods are much more numerous than
the cones, though towards the periphery of the retina

THE EYE 13

the cones become more numerous. The total number of
cones has been calculated as being more than three
millions.

The macula lutea, or yellow spot, is oval, subtending
an angle of about 6° in its longest axis, and 4"^ in the
shortest. As its more common name indicates, it is
distinguished from the rest of the retina by its yellowish
brown colour. At its edges, the oval is slightly thickened,
but in the centre it becomes very thin, and is there termed
the fovea centralis, and is about '3 mm. in diameter.
The general character of the layers in the macula are
pretty much the same as in the rest of the retina, except
that the cones are more numerous than outside. The
rods diminish in number as they approach the fovea
centralis, where they are altogether absent. The colour
is due to a pigment staining one or more of the layerSj
but is said to be absent at the fovea. The yellow spot
we shall have to take into account in our experiments.
It is a continual source of difficulty in making measures,
more particularly as the amount of colouring matter
often varies in different observers.

Zone of Distinct Vision.

Another feature of the optical arrangements and the
retina is that there is a zone of distinct vision near and
around the axis of the eye. When we look at a small
point of light, such as a star, the image falls on the
fovea centralis. When two points of light subtend an
angle less than about one minute of arc, the two images
are blended one into the other, and the separation is not
noticed. Calculating the distance apart on the retina
on which the two images fall, it is found that they are
about xi^ <^f ft millimetre apart. In the human eye the

14 RESEARCHES IN COLOUR VISION

distance apart of the cones are about xx^n ^^ ^ milli-
metre from centre to centre, and the diameter of a cone
is somewhere about xiAr^ ^^ ^ millimetre. Hence if the
images of two points of light are about xtfW ^™- apart,
they both may fall on the same cone, and this would
account for only one sensation being stimulated ; but ^ve
also have to remember that optically the image of a
point is a disc of definite size dependent on the length of
focus of the lens ; and if the two discs overlap, no separa-
tion would be apparent. Away from the central spot in
the retina the distance apart of the objects has to be
gradually increased in order to eflfect separation. From
a study of the optics of the eye, this appears to be due to
the aberration of the lens and the curvature of the surface
on which the images are received. The rays become
oblique and have to follow the ordinary laws which
govern the definition given by such rays. It may be
said that distinct vision is really confined to a circle
subtending about 3^.

The lens of the eye is not an achromate, as there
is no correction possible in its construction. We have
only to place a piece of violet gelatine tissue against
a small hole in a card and look at a luminous point.
Violet, such as methyl violet, only allows the red and
blue to pass, and cuts off the intermediate rays of
the spectrum. In one case the red in the violet will
give a sharply defined point with a blue halo round it,
and in another case there will be a blue point surrounded
by a red halo. The spectrum itself, when looked at
from a distance, will also indicate the want of achro-
matism. The violet end will not be in focus when the red
appears sharply defined. There are other defects in
some eyes that may be encountered, and which we shall
allude to in a subsequent chapter, as also the character

THE EYE 15

of the sensations which are stimulated by the impact of
light on the retina.

So far, then, we have described the human eye, and
this, with some slight alteration in the description, will
answer for, at all events, most mammalians.

Seat of Visual Impulses.

We must next indicate where the visual impulses
are originated.

If we take a candle or a lamp into a dark room in
which a large piece of grey or white paper is hung on
the wall, and illuminate the paper with it, holding it
close by the side of the head, there will appear in the
field of vision of the eye nearest the lamp an image
of the blood-vessels of the retina. The light enters the
eye through the cornea, and an image of the lamp or
candle-flame is formed on the nasal side of the retina.
The light coming from this image throws shadows of the
retinal vessels on to the other parts of the retina. (This
explains how it is possible to see these vessels oneself.)
The same effect is seen when a second

C F

person concentrates the rays by a
low-power lens on to the sclerotic
coat near the cornea. The light by
this plan comes to a small point on the
sclerotic, and passing direct through
the vitreous humour, casts a direct
shadow of the vessels on to the retina.
Thus, taking one vessel, S, the
light when concentrated at A casts ^iq, 3.

a shadow on the retina at B, which
shadow is seen projected along BC. If the concentrated
light is moved to D, the same vessel casts a shadow

16 RESEAKCHES IN COLOUR VISION

at E, and this shadow is seen projected along EF.
Knowing the nodal point of the lens and its distance
from the points C and F, as seen on the screen, and the
distances apart of C and F, as the other distances are
also known from the diagrammatic eye (Fig. 2), it is easy
to calculate the distance of S from the sensitive layer of
the retina. The calculation shows the sensitive layer to
be that in which lie the rods and cones — i.e, the retinal
layer which is farthest from the lens and nearest to the
black pigment layer.

This is an important fact in colour measurement,
for it shows that rays of light falling on the macula
lutea have to pass through the yellow pigmented layers
before they reach this sensitive layer. In other words,
partial absorption of violet and blue rays takes place
before the sensation is stimulated in the rods and cones.
In the portions of the retina outside the macula prac-
tically no absorption has to be taken into account. The
reader should have this well in his mind.

The Blind Spot.

One other spot in the retina must be mentioned,
viz. the blind spot. It is that spot where the retina
is directly connected with the optic nerve.

The blind spot can be readily shown to exist by
making on white paper two dots about 4 inches apart,
fixing one eye (the other being closed) on the left-hand
dot (if the right eye be used), and moving the paper
to and fro from the eye. At one distance the right-hand
dot will disappear, but reappears when the paper is
moved nearer or farther from the eye.

THE EYE 17

Evolution of the Bye.

The eye, as described, appears to have obeyed the
laws of evolution, since in the living creatures which
now exist we have evidence of a primitive eye gradually
becoming more perfected until we arrive at the mam-
malian eye.

The most primitive organ of sight, we are told, is
perhaps to be found in a snail. There is apparently
in it no organ of vision, and yet it feels the light.
Examination shows that there is a thinning of the
skin on each side of the head ; and when the creature
is exposed to light and darkness alternately, the move-
ments of the body show that it has a sensation of light.
Another very primitive eye
is that of the nautilus. Here
we have a depression in the
head, shown in the right-
hand figure. Its organ of
vision is in fact a pin-hole
camera, the pin-hole being ^^^' ^'

large size. Photographs taken with such a camera
would be most ill defined ; and there is not complicated
retinal structure which would indicate that these
creatures would have any sense of colour. The objects
it would see would be probably black and white, and
their definition would be of the worst character.

Again, we have other creatures where there is
evidence of an embryo lens, which fills up the space
of the pin-hole. The retinal structure is said not to
be indicative of an apparatus for receiving the impres-
sion of colour. If so, we have an eye which is adapted
for monochromatic vision, with definition of form far
superior to that of the nautilus.

B

18 RESEARCHES IN COLOUR VISION

We can examine the eyes of other creatures, and
can find amongst them some in which there is a better
formed lens, and a rudimentary iris.

Continuing the examination, we find further im-
provements in the optical and retinal parts of the
eye, till we arrive at that eye we have tried to describe
at some length.

In our own eyes there is evidence that the colour
sense has been evolved, and a very simple experiment
carried out by the reader will convince him of two
things — first, that the sensation of light exists quite
regardless of colour, and that the two do exist together.
Let us place a spot of green on a black surface either by-
throwing a spot of spectrum green on to a white surface
in a dark room, or else let us place a green wafer on a
sheet of black paper in a well-lighted room. Standing
some feet away from the spot, let one eye be closed and
the spot be viewed with the other eye in the ordinary
manner. The green spot will be seen and the image
will fall on the centre of the retina. Next turn the
head and eye together, so that the image of the spot
falls on a portion of the retina approaching the peri-
phery. The image of the green spot will at a certain
distance from the axis and beyond become pure white.
The green colour will have disappeared entirely, and no
notion of the hue would be formed if the image of the
spot was thus received without having been seen on the
centre of the retina.

A simple plan, and one often practised, is to mount
a green or other coloured wafer on the end of a thin
rod such as a long pencil, and to obtain the assistance
of a second person to help in making the experiment.
The experimenter fixes his eye on that of the assistant,

THE EYE 19

who holds the wafer at the distance of distinct vision
and near his nose. The eye and the wafer are seen
by the experimenter. The assistant gradually moves
the wafer away from the nose towards the right or
the left. The experimenter keeps his eye fixed on
that of the assistant, and at some angle which the line
joining the wafer and the experimenter s eye makes
with the line joining the eyes of the two parties, the
wafer appears colourless. The angle made is often
guessed.

The disappearance of colour from wafers of other
hues can be noticed in the same way. It must be
stated here, though it will be restated farther on, that
the brightness of the colour and the size of spot cause
variations of the angle at which the colour disappears.
In fact, if the brightness be feeble and the angle which
the coloured disc subtends on the retina be very small,
a shift in the axis of the eye of a very few degrees will
suflBce to render the spot colourless.

This simple experiment is well worth consideration,
as it shows that the retina is most sensitive to colour
round the part which the axis of the eye cuts, and that
it gradually diminishes in sensitiveness to colour, though
not necessarily to light, as the periphery is approached.
This is exactly what one would expect if the eye has
obeyed the laws of evolution, and it has to be reckoned
with in certain measurements of colour which are to be
described. It may be stated here that every individual
is colour blind, though not light blind, in the outer part
of the retina. The most diflBcult colour to cause to
disappear is the blue, and it gives us the hint as to
which colour was the first to be evolved. The sensa-
tion of light is shown to be white, and colour has been

20 RESEARCHES IN COLOUR VISION

added on to the sensation, or a portion of that sensation
has been converted into colour. We shall see farther
on that apparently there are three distinct visual
colour apparatus and that light is the fundamental
sensation of each colour sensation.

N

CHAPTEK III

ON PHENOMENA IN VISION

It may be as well to mention briefly some facts con-
nected with the visual effects of the impacts of light,
coloured or otherwise, on the retina, which occur and
are often overlooked. Those given here are taken from
the late Mr. Shelford Bidwell's papers, printed in full in
Nature and in the Proceedings of the Royal Society.
They are also epitomised in a small book called Curiosities
of Light and Sight, ^ He brought together and ex-
panded some of Charpentiers admirable work, adding
his own experiments, and giving explanations of pheno-
mena which appeared to require elucidation.

Recurrent Vision.

When a flash of white light is received on the retina,
we have what are called positive recurrent optic images.
Bidwell tells us that they were first accidentally dis-
covered by Professor W. Young, when experimenting
with a large electrical machine. Young noticed that
after a strong spark had muminated any conspicuous
object, it was seen at least twice, the second time after
an interval slightly less than a quarter of a second after
the first — the first image was vivid, the second faint.
Often it was seen a third time, and sometimes even a
fourth time. This phenomenon he called "recurrent

^ Messrs. Allen & Co., to whom we are indebted for permission to use the
iUustrations in this chapter.

21

^

y

k

22 RESEARCHES IN COLOUR VISION

vision." Bid well gave it the name of the Young effect.
Let one pole of an influence electrical machine be con-
nected with the inner coating of a half-pint Leydeu jar
and the other with the outer coating, and the dis-
charging balls be placed one-quarter inch apart, and a
white object such as a small sheet of paper be placed
in an upright position a few inches away from the
terminals of the machine, and then the machine be
worked till the discharge takes place, the room being
darkened. If the eyes are screened from the sparks and
are directed towards the white object, the object will
be seen, and in about one-fifth of a second a recurrent
image will make its appearance, and after another .
interval of darkness a second faint one will often be
seen. Bid well says that under favourable conditions »
he has observed as many as six or seven reappearances
of an object illuminated by a single discharge. It
must be recollected that the light of a discharge is I
excessively intense, and that owing to its very short
duration the retina does not fully realise how intense
it is.

The Ghost.

This recurrent image can be well shown to an
audience on a screen by one of Bidwell's many devices.
A metal disc of some 2\ inches diameter, which can
be turned about its centre, is prepared and placed in
position as a lantern slide. A small circular hole is
drilled near the periphery. When the disc is focused
on the screen, we have, of course, a dark image of
the black disc, but a small spot of white light near
the rim. When the disc is slowly rotated so that the
spot travels round the screen with merely a slight
elongation in the line of its travel, if the eyes be kept

ON PHENOMENA IN VISION 23

steadily fixed on the screen, it will be found that a faint
violet spot travels behind the -white oval, separated by
an interval of darkness. When
the speed of rotation is increased,
the interval between the two spots
will increase, and when the rota-
tion begins to stop the two spots
come nearer to one another, and
finally the two merge into one
another. If a green glass be
placed in the beam, the violet of
the ghost of the green spot will ^^^ g

become apparently more intense.

With orange glEiss the intensity of the violet becomes
less ; whilst if the light be made red, there will be no
ghost to the red spot.

Bidwell made an apparatus by which he could repeat
this experiment with a spectrum colour, and found
that the one colour which gave no ghost was the red.
When the whole spectrum, as a line, was rotated, the
ghost to each colour, except the red, was violet. The
time of rotation being known, and the interval between
the original spot and the ghost being measxu^d, we
have the means of calculating the interval that elapses
between the first image and that caused by recurrent
vision, which Bidwell puts down as closely one-fifth
of a second. The interval of time which elapses between
the two images seems to be the same for all colours.
If there were a difference this accurate observer would
have noted it.

Charpentier's Law.

Charpentier, the eminent French scientist, made
many observations on the impressions received on the

24 RESEARCHES IN COLOUR VISION

retina by light. Charpentier's law, BJdwell tells us,
is this : " When darkness is succeeded by light, the
stimulus which the retina JSrst receives, and which
causes the sensation of luminosity, is followed by a brief
period of insensibility, resulting in the sensation of
momentary darkness. It appears that the dark period
begins about one-sixtieth of a second after the light
has first been admitted to the eye, and lasts for about
an equal time. The whole alternation from light to
darkness and hack again to light is performed so rapidly
that, except under certain conditions, which, however,
occur frequently enough, it cannot be detected."

The apparatus which Charpentier employed for
demonstrating and measuring the duration of this
effect is very simple.

"It consists of a blackened
. disc with a white sector mounted

/ upon an axis. When the disc

I is illuminated by sunlight and

\ turned rather slowly, the direc-

tion of the gaze being fixed upon
the centre, there appears upon
p.jg g the white sector, close behind

its leading edge, a narrow but
quite conspicuous dark band. The portion of the retina
which at any moment is apparently occupied by the
dark band, is that upon which the light, reflected by
the leading edge of the white sector, impinged one-
sixtieth of a second previously."

Bidwell, with a 4-inch disc of black having a slit
of about one-fiftieth of an inch wide at the circumference,
placed in front of an illuminated ground glass, was able
to show more of these dark periods. When the eyes
were fixed on the centre, and the revolution of the disc

ON PHENOMENA I

was about once a Becond, the disc
ance as given in the figure.
Charpentier effect occurs at the
beginning of the period of illu-
mination, and a dark reaction at
the end of the period of illumina-
tion. He also explains and shows
that owing to what is called the
proper light of the retina, or what
we call the intrinsic light, ordi-
nary darkness is not an absolute
black, and says that the darkness
which is experienced after the
extinction of a light is for a sma
more intense than common dar
experiments to show this abnormi
he gives a diagram of the differei;

ll

Ziflu On

Fluctuattent

auady

during the continuance of light £
till the first recurrent image is see
that to the next recurrent image 1
and is that of the intrinsic light ol

26 RESEARCHES IN COLOUR VISION

Coloured Borders to Black Lines.

There is another phenomenon connected with the
visual effects which ought to be noticed, of which the
origin has been traced by Bidwell. It is not uncommon
for a reader to go to sleep with a book in his hands.
He may wake suddenly and turn his eyes at once to his
book, when he will find that the printed matter instead
of being black becomes red, and it is only after several
seconds that the black of the printed matter is seen.

Amongst various other phenomena, Shelford Bidwell
traced the cause of this one, and communicated his
observation to the Royal Society^ in 1896 and 1897,
and the following description is abstracted from his com-
munication. After several preliminary experiments, he
describes one in which a white card disc with a diameter
of 6 inches was employed. A sector of 60° was cut out,
and the remainder of the disc was divided into two
equal parts by a straight line from the centre to
the circumference, and one of these parts was painted
black. The disc was attached to a horizontal spindle,
and was rotated five or six revolutions per second, whilst
its front was illuminated by a lamp of sixteen-candle
power. A white card, on which was a black line or design
composed of black lines was supported behind the disc,
and viewed intermittently through the open sector.
When the rotation was such that the open sector
succeeded the black part of the disc, and was then
succeeded by the white portion, the black lines were
received as red.

When other experiments were carried out, it was
found that a bluish-green border became visible when
the illumination was increased, and that with a still

* proceedings of the Royal Society, vols. Ix. and Ixi.

ON PHENOMENA IN VISION 27

stronger illumination the red was entirely replaced by
the greenish blue. It may be stated that if the lines
are wider than about ^^5 of an inch, when observed at
a distance of 2 or 3 feet, as the thickness increases, a
black central line is seen bordered in red, the borders
lying in the black. It is only when the lines are thinned
down that the coloured borders meet and cover the whole
of the lines.

When the sector was rotated, so that the black
part followed the aperture, there was no suspicion of
red, the lines appeared to become blue. This appear-
ance, Bid well states, is partly if not altogether illusory.
It is the bright ground in the immediate neighbour-
hood of the lines that l)ecomes blue, the lines them-
selves, except possibly just within the extreme edges,
become grey, owing to the alternation of black and
white. When a small card was placed behind the
rotating disc, it merely turned a grey, without any
suspicion of blue in it. From other experiments,
Bidwell appears to show conclusively that the red
effect on the black lines is due to a spreading sym-
pathetic action for a short interval of the red-
perceiving apparatus, when the retina receives the
impact of white light after a period of darkness, and
that coloured light in which no red was present gave
no colour to the lines. The green-blue lines which
succeed the red when strong illumination is given to
the card is the after-image of the red (see Chapter
XXIII.). The blue border outside the lines, which is
seen when the aperture succeeds to the white, is
probably the after-effect of the red. We thus see
that in considering the effects of light on the retina,
account has to be taken of its duration and of the
state of sensitiveness due to darkness.

28 RESEARCHES IN COLOUR VISION

J^^ Bcnkam's Top.

,yV' "^ One of the objects of Bidwell's investigations was
^ to account for the colour phenomena which are pro-

i^ duced in Benham's spectrum top. The figure below

gives the idea of the
top, which is a disc of
about the size of the
figure.

When the disc is
rotated rather slowly
round its centre, each
group of black lines
will probably appear of
a different colour. The
hue depends on the
■ speed of rotation and
on the brightness of
the light. The maker
of the top says that
when the top is rotated
in the direction of the
arrow the outside set of lines will appear red and the
inside one dark blue, whilst the intermediate lines
will show a green colour. It will be noticed that the
rotation gives black first, then the outside lin^ on
a white ground, then the second and third set sand-
wiched in between the outside and inside lines, which
latter end in the black background. On quiet rotation
the hues appear as above, and it wiU be seen that
Bidwell's experiments confirm the idea that the red
of the outside Hues is due to a spreading of the red
sensation excited by the white to neighbouring portions
of the retina, on which the lines are received. If the

ON PHENOMENA IN VISION 29

lines are made thick, the red borders are seen along-
side a black central space. The blue seen in the
inside lines apparently is a mixture of the black
background and of the green-blue band which is seen
when a straight or curved edge of a white surface is
darkened (p. 27). Such an edge shows a blue-green
border for about one-fifth of a second, owing to
"a sympathetic insensitive reaction" in the receiving
apparatus outside the image. The writer made ex-
periments with this same top, illuminating it with the
white arc light and with monochromatic patches of
light, and several observers gave the same descriptions
of what they saw when the disc rotated before them.
Calling the outside set of lines No. 1, the next No. 2,
the next No. 3, and the innermost No. 4, when the
rotation caused the lines to affect the retina after a
period of rest by the black, the effect produced by
moderately luminous white was —

No. 1, crimson,

No. 2, olive green,

No. 3, grey (slightly violet).

No. 4, dark violet.

When the illumination was by red light (C of the
spectrum) —

No. 1 was red,

No. 2 was lighter red,

No. 3 was very light olive green.

No. 4 was darker olive green.

In this light no other sensation but red and a very
little green was in the colour used, which probably
accounts for the colours in Nos. 3 and 4, for when the

J

30 RESEARCHES IN COLOUR VISION

red was that of the lithium hue, in which there is
only the red sensation stimulated —

Nos. 1, 2, 3, and 4 all were red.

When a green light close to the green magnesium line
of the spectrum was the illuminating colour —

No. 1 was bluish green,
ji No. 2 was lighter bluish green,

^ ^ No. 3 was same as No. 2,

' • .\/ No. 4 was ruddy black.

^ V 1^1 this colour the constituents may be said to be the

green sensation and white (see Chapter XV.).

When the blue of the blue lithium hue was the
illuminaut —

No. 1 was grass green,

No. 2 was lighter grass green.

No. 3 was same as No. 2,

No. 4 was ruddy black.

In this case we have the blue sensation mixed with
a large quantity of white (see Chapter XV.).

When the illuminant was the whole of the violet
of the spectrum —

Nos. 1, 2, and 3 appeared light violet, and
No. 4 darker violet.

In the case of the violet there are only the red and
blue sensations present (see p. 240).

The two next experiments are interesting, as the
illumination was with white light, but the white was
compounded of two rays only, two slits being opened
in the spectrum and white matched.

ON PHENOMENA IN VISION 31

The illuminant was the white of a mixture of red
and bluish green : —

No. 1 was indigo blue,

No. 2 was reddish orange,

No. 3 was same as No. 2,

No. 4 was darker orange.

The next illuminant was the white of a mixture
of spectral yellow and blue : —

No. 1 was sky blue,

No. 2 was sage green,

No. 3 was same as No. 2,

No. 4 was bluish black (perhaps black).

These phenomena are explained if we take it that
each sensation has its own sympathetic action on the
sensation-receiving apparatus, that of the red being
greatest and that of the blue least.

Some experiments recorded in Nature by Finnigan
and Moore with broad lines fully bear out Bidwell's
contention as to the colours seen in the lines when
illuminated by the arc light. They made the lines a
centimetre broad and found that on rotation the band
following the black was bordered with a red over the
black, and on that which came from white to black the
band was bordered on the white with a blue to green
colour, leaving the band quite black. Bidwell's explana-
tion, as before said, is that the red colour of the
fine lines following the black are due to sympathetic
spreading of the red sensation, whilst the blue colour
of the fine lines following the white is due to the want
of such sympathetic action when the illumination is
rapidly shut off, leaving the other sensations exhibited
on the black surface on which those lines are practi-

32 RESEARCHES IN COLOUR VISION

cally viewed, and which the retina takes as part of
the lines, though the colour is outside their border.

When the speed of rotation of the disc is gradually
increased, the red of the outer lines grows darker
and duller, and then, passing through a transition,
which Messrs. Finnigan and Moore were unable to
observe, the lines assumed a vivid green, then a blue,
and when the rotation was very rapid they assumed
a violet hue.

In regard to Benham's top, it has been stated
that the colour phenomena were due to the want of

achromatism in the eye. Mr.
Bidwell conclusively showed that
it was not.

He quotes an experiment
which is fatal to the theory.
He prepared a disc as above,
and spun it above *a page of
printing. The letters beneath
that part of the disc that is
partly white and partly black
will appear red, but those beneath the remainder will
always appear black. As he remarks : *' The demar-
cation is quite definite, and a single printed word may
be made to appear red in the middle and black at its
two ends." It is, of course, impossible that the lenses
of the eyes should be perfectly accommodated for the
letters which appear black, and, at the same time,
seriously out of focus for the others.

CHAPTEE IV

COLOUR PATCH APPARATUS

Colours which we see round us are almost invariably
impure colours — ^that is, colours which on analysis are
found to be composed of mixtures of pure colours, or to
be pure colours mixed with white. Thus the yellow of
the marigold, which is a brilliant orange yellow, although
it appears to be a pure saturated colour, is found to be
composed of all the spectrum colours, from red to yellow-
green, which have certain relative brightnesses to one
another which differ from those found in white light.
It is proposed in this chapter to describe an apparatus
which can be used for the quantitative measurement of
brightness and of certain qualities which the spectrum
colours possess, and at the same time to show that it
is equally useful for the measurement and analysis of
colours which are seen in such objects as the marigold.
With this purpose in view, we must have an apparatus
which, when applied to the spectrum, shall not only be
able to isolate a slice of colour from the spectrum by
a slit placed in it, but also to produce a patch, at least
li inch square, of the colour which passes through the
slit or a mixture of the colours which pass through more
than one slit.

It would be foreign to our purpose if we described
in detail the spectroscope as ordinarily used. It is sup-
posed that the reader is familiar with its principles, and
any description that may be given here will only be

33 . ^

34 KESE ARCHES IN COLOUR VISION

such as is necessary to understand the lines on which
the " colour patch apparatus " was designed.^

ColliTnator.

In the spectroscope, which is dependent on prisms
for the dispersion of light, there is a slit at one end
of a tube with a lens at the other end, to render the
rays coming through the slit parallel. The lenia is there-
fore of such a focal length that the slit is at its equiva-
lent focus. The slit, tube, and lens form what is called
the collimator. The slit can be closed or opened by
a screw arrangement ; and here it should be remarked
that for exactness of measures the jaws of the slit shovZd
both move through equal distances outwards or inwards,
so that the line of junction of the jaws when closed
should always be the central line of any aperture to
which the slit is opened. The necessity of this will
be apparent when it is called to mind that every colour
in the spectrum, when focused, is an image of the slit,
and that the central line of the slit is the centre of the
coloured image. Should one of the jaws be fixed whilst
the other is movable, the centre line of the slit moves
from the line of junction when the slits are closed
through half the width of the slit, and this entails a
corresponding movement of the coloured image of the
slit. As the rays of the light falling on the slit emerge
from the lens as parallel rays, they will fall on the sur-
face of the first prism as parallel rays, and all the rays
of each colour will have the same deviation as they pass
through it, and through as many prisms as are placed in
its path. The deviation alters in amount according to
the angle at which the surface of the prism is placed in

^ See Papers Nos. 4, 5, and 6.

COLOUR PATCH APPARATUS 35

reference to the axis of the parallel beam. For con-
venience' sake, it is better to have the surfaces of the
prisms so placed that the central ray of the spectrum
(say), the blue-green, shall have what is called "mini-
mum deviation."

This angle of minimum deviation is readily found
by throwing a defined spectrum formed by one of the
prisms on a screen. The ray selected is watched whilst
the prism is turned on its base right and left. One
position will be found where the selected ray seems
to have no movement ; though turning it either to right
or left, the ray will commence to travel along the screen
in the same direction. The angle, when the motion
ceases, is the angle sought for. It must be remembered
that each ray has its own angle of minimum deviation,
and the blue-green ray is chosen for convenience, as
dividing the spectrum into fairly equal parts. When
the first prism has been fixed, a second and a third may
be placed in the path of the beam, and the angle of
minimum deviation found with the added prisms in
the same way. Care must be taken that the slit is
parallel to the edges of the prism, otherwise a vertical
line of colour in the spectrum may not be of the same
hue throughout.^ The surfaces of the prisms should be
accurately vertical, and usually this can be done by
levelling the bases. Our own prisms are very colour-
less and made of medium flint-glass. Two of such
prisms give a dispersion which is quite sufficient to form
a spectrum some 3^ inches long. The "angle "of the
medium flint prisms we use is 62^°, their height 1^
inch, and the width of face 2 inches.

The collimator tube is for steadiness supported on a
stand of nearly its own length, and rests on two Vs.

^ There is always a very slight curvature of bright lineis in the spectrutD.

36 RESEARCHES IN COLOUR VISION

The collimator and prisms are each supported separately,
the former being rigidly fixed, so that there is no
" spring " to it (which is not usually the case in spectro-
scopes found in a chemical laboratory). It is most
important that a collimator should be rigidly fixed in
regard to the surface of the prism.

The prisms are mounted on separate brass bases
with levelling screws, to secure that their faces can
be made truly vertical ; and when the angle of minimum
deviation for the central ray has been found, the brass
levelling screws find a bearing in the brass plate below,
in which depressions are made in positions corresponding
to this angle.

For forming an image of the spectrum an achromatic
lens of 30-in. focus is employed. It is mounted on a
camera which has a rack and pinion focusing arrange-
ment. The focusing screen has a horizontal swing-
back, which allows one end of the spectrum to be at
a longer focal distance than the other. This is necessary,
as the focal distance from the lens for the violet is
shorter than that of the green, and still shorter than
that of the red. There is the usual ground-glass screen
for focusing, and grooves which take dark slides hold-
ing plates 6J by 3J inches. The instrument as now
described is a photographic spectroscopic apparatus.

Slide in the Spectrum,

In place of a photographic plate, the grooves will
take a metal or wooden slide, in which is inserted a
brass panel and slits, as will be found described farther
on (p. 41).

If we remove both ground-glass and slide, and, a
short distance in front of the position where the spec-

COLOUR PATCH APPARATUS 37

trum is in focus, place a lens of some 4J^ to 6 inches
in diameter and having a focal length of about 3 feet, a
white image of the face of the first surface of the first
prism can be obtained on a screen some 4 to 5 feet
distant from the focusing screen. The lens recombines
the whole of the spectrum if its axis makes a slight
angle with the direction of the central ray.

We can apply the optical formula F = fi'!fi_ , , where

F is the focal length of the combination of the lenses
which form the spectrum and the '* collecting " lens.
Let the TSiysf^ and/, the two focal lengths, be 30 and 36
inches respectively, and s (the separation of the lenses),
about 36 inches, we find that the focal length F is 3 feet
and the optical centre is about half-way between the two
lenses, or 1^ foot from the first lens. The first surface
of the first prism may be taken to be 2 feet away from
the optical centre, so that the sharp image of the surface
of the prism will be about 6 feet from the optical centre,
or 4 to 6 feet from the combining lens. If the axes
of the two lenses lay in a straight line, the image would
be bounded by fringes of colour, but by causing the axis
of the combining lens to make a slight angle with that
of the first lens, the fringes can be made to disappear.
If now the wooden slide, with a slit located in the brass
fitting, be passed through the spectrum, the image of the
surface of the prism will be found to be of the colour
which passes through the slit, so that a monochromatic
patch of light of any colour can be thrown on the screen.
Instead of a screen, it is convenient to have a cube
covered with white material mounted on a rod and
backed by black velvet, on which the patches shall fall.
This isolates the patch, and the colour is backed by
a black ground. Even if the slit on the spectrum be

38 KESEARCHES IN COLOUR VISION

opened wide, the colour will still be practically mono-
chromatic, since it is found that the rays on each side of
the central ray passing through the slit, when combined,
match it in colour. If the recombining lens be removed,
there will still be coloured patches showing on the
screen, but as the slit is moved from red to violet there
will be a continuous travelling of the patches along the
screen. The recombining lens keeps the patches in the
same place.

Single Colour Patch Apparatus.

The above general remarks show on what principle
the colour patch apparatus was constructed, and the
next figure shows it as it at present exists.

In this apparatus only one colour patch can be
formed. The rays R, R, coming from the crater of the
positive pole of the electric light, are collected by a
lens Lp and an image of the crater thrown on the slit S^.
After passing through the collimator C, the rays emerge
as parallel rays ; part passes through the prisms P^ and
Pg, and is collected by a lens, Lg, of about 30-inch focal
length, and a spectrum is formed on a focusing screen at
D, which is removed, and a slide inserted in which slits
can be placed. The image of the surface of the first
prism is formed on the white surface of a cube, E, by
means of the lens L^ (of about 30-inch focal length), so
arranged that the image of one edge of the prism falls
at a, the other edge falling outside d. Part of the beam
which passes through the collimator is reflected from
the surface of the first prism to a mirror G', and passes
through a lens, L^, then through a bimdle of glass, G° ^
placed at an angle to the beam, and on to the surface

1 For ordinary work the bundle of glasses G" is not required/ which does
away with the mirror G"* and the sector M".

COLOUR PATCH APPARATUS 39

dc of the cube, a rod, Kp being placed in its path, to
secure that this white beam does not fall on ad^ on
which the colour mixture falls. The portion of the
beam which is reflected from G" is again reflected by

/I

f

%

i^

„-

4C

r^"'

/

;> /*'

/; ^

G"', a silvered mirror, on to cd, a rod, Kj, placed in its
path prevents it falling on ad or ac as desired. In all
three beams, sectors, M', M", and M"', can be placed, to
allow any or all to be reduced in intensity at pleasure. In
the beams X and Y any absorbing medium desired can

40 RESEARCHES IN COLOUR VISION

be placed. A small ray of
light, Z, is allowed to pass
beyond P„, and falls on &
small mirror, G"', which
reflects it on to the back
of D, casting a shadow of
a needle, N {fixed to B,
the camera), on a scale
at the back of D.'

L^ is a lens of short
focus which can be moved
into a fixed position be-
hind L^ to throw an en-
larged image of the slit on
a scale placed above dc.

SHts and Slit Holder.

There is one part of
the apparatus which must
be shown in some detail,
viz. the slide D and the
slit holder. The slide D
is shown in the annexed
figure.

In Fig. 12 F is a
brass plate with necessary

' In the writer's present appa-
ratus the needle is done away with
and a tmiiapart^nt scale is mounted
in the top o[ D, and a small lens in
front of the scale tlirowa a magni-
fied image of the graduation on a
distant screen. {See the descrip-
tion o( the modified apparatus,
p. 45, tor detnJis.)

COLOUR PATCH APPAftA-TUS 41

grooves cut In it (see Fig. 13). A, B, and C are three
slits wliich can be clamped in any position by means
of the screw GG. H is a slit which is always kept in
one position and has a fixed and carefully
measured opening {used for measures to be
compared together from time to time). (There
is a transparent scale S fitted into D, through
which a beam of light passes on to a distant
screen with a mark on it ; see p. 45.) XX are
two grooves cut the whole length of the top
and bottom l)ars, as also are YY. In XX the
slits (of which a full-size figure of one is shown ^
in Fig. 14) slide along XX and thin black cards (.,„ jg
EEin YY, Fig. 12.

The slits in the brass frame, it will be seen, are made
to open centrally, so that the centre line of any aperture
is always in one position.

The brass frame F is fixed to a
bard-wood slide D (in which there is
a rather smaller opening than the
dimensions of the brass plate).

This apparatus is all that is, as a
rule, required for colour measurement
and mixture.

[It may here be noted th&t by re-
moving the slide at D, and then placing
a lens of 9- or 10-inch focus ^me 4
inches in front of the recombiniiig lens,
an enlarged spectrum can be obtained
on a white screen placed at the same
distance as the cube.]
The lens L', which throws an image of the crater
of the arc, should have such a focal length that the
length of the slit is well covered by the brightest part.

42 RESEARCHES IN COLOUR VISION

It should also be of such a diameter that the ratio
of diameter to focal length is not less than the ratio
of the diameter of the collimating lens to its focal
length. If it be less, the collimating lens will not be
filled with light. [It should be noted that the smaller
the ratio of focal length to diameter of the collimating
lens the brighter will be the spectrum.] If Fig. 11 be
examined, it will be seen that any variation in the
brightness of the spectrum is accompanied by a corre-
sponding variation in the light reflected from the first
surface of the prism. This is a most valuable property,
as the brightness of any colour is most frequently re-
ferred to in terms of the brightness of the reflected
white beam.

Apparatus for using two Spectra simultaneously.

A later form of colour patch apparatus* is arranged to
enable two spectra formed by the same source of light to
be used either separately or together. This arrangement
allows a comparison of any differing mixtures of spectrum
colours to be made, and it also allows the addition of
any desired quantity of white light to the colour patches
formed by the aid of either of the two spectra.

In this apparatus, as in the last, the intensity of the
white light used for comparison with the colours varies
with the intensity of the spectrum. The same white
light is used as 'before to form the spectrum and the
reflected white light as the comparison light, but, in
addition, the main light, after passing through the two
prisms, passes through a half-silvered mirror, inclined at
about 45*^ to the axis of the lens. The rays reflected are
again reflected so as to pursue a course roughly parallel

1 See Paper No. a

COLOUR PATCH APPARATUS 43

to the main spectrum. Thus two similar spectra are
placed side by side. The accompanying diagram will
show the arrangement.

As in the apparatus described, E is the source of
light used outside a darkened room, Lj, L^ are lenses
throwing an image of the source of light on the slit Sj
of the collimator C. The parallel beam passes through
the prisms P^, Pg and is received on a colour-corrected
photographic lens, L^, of suflBcient diameter to take in
the whole of the light coming through the prisms.

The lens forms a spectrum on a focusing screen at
Dp which can be removed and slits Sg placed in the
image. L^ collects the colours and gives an image of
the face of the prism P^ on the screen B.

Behind the lens L4 is placed the semi-silvered mirror
Mj, reflecting, as nearly as may be, the same amount of
light as is transmitted through it. If the mirror be on
a plate of glass with parallel sides, it should be as thin
as possible, to avoid any serious mixture of colour in the
second spectrum due to the reflection of the unsilvered
surface. If a plate be made up of q
two thin prisms, as in margin, with
the surface AB of one of them half
silvered, the transmitted beam is ^
not deviated, and the beams reflected from DB and AC ^
are diverted and not used.

The reflection from the semi-silvered mirror M^ falls
on a silvered mirror, Mg, which reflects the beam in such
a direction that it falls on B, the image of the spectrum
being thrown on Dg. The image of P^ is thrown on B
by the lens L^. A beam of white light is reflected from

* The two thin prisms are used in order to protect the silvered surface.
One thin prism by itself may be employed, but the length of the direct
spectrum will be slightly increased or diminished according to the position
of the thin end of the wedge.

44 BESEAECHES IN COLOUR VISION

s* V-'

\ •'■

it^:

y

COLOUR PATCH APPARATUS 45

the face of P^ by Mg (which may be either a silvered
mirror or plain), and is also focused on B, so that we
have the patches from both spectra and from the white
light falling over one another on B. By means of rods
correctly placed, a colour or colours from either spectrum
can be isolated and be mixed with any proportion of
white by using sectors as shown. There are slides
carrying the slits at D^ and Dg, and to them are attached
transparent scales. In the case of D^ a beam of white
light falls on the mirror M^, as shown, and passes
through the transparent scale at a, and a lens X throws
a magnified image of the graduation on a distant white
screen, on which a zero mark is drawn. This enables
the transparent half-millimetre scale to be read to a
tenth of that unit. In a similar way the scale at a
is magnified by X' by a beam of light falling on M^.
When the scale readings are not required, the sources
of light illuminating them are covered up.

Again the lenses A^ and A^ are mounted in a sliding
arrangement and can be moved in firont of lenses Lg and
Lg. When a slit is drawn in front of A^ or A^ the
image of the aperture is magnified on a distant screen,
carrying a scale, and the width of the slits can be
accurately ascertained by noting on such scale the
reading of the breadth (say) of \ millimetre width of slit.

Still more recently the apparatus has been altered in
one particular. The half-silvered mirror M is replaced
by a fully silvered mirror or a right-angled prism, which
reaches to half the height of the prism. The bottom
half of the beam is totally reflected to Mg, and a spectrum
is as before formed at Dg. On reaching the screen B,
each patch is half the height of the full patch. By this
means any difficulty about half-silvering is avoided, the
slight second spectrum which overlaps the main spectrum

46 RESEAKCHES IN COLOUR VISION

from the reflection from the back of the semi-silvered
mirror of plane glass is entirely absent. Further, the
two spectra are very nearly equally bright.

The Receiving Surface.

In early experiments that were made, white cardboard
was used as a receiving screen, and for ordinary work
answers well ; but the question arose as to whether card
of the same kind of whiteness could always be obtained.
This led to the conclusion that a white of definite
"whiteness" ought to be used. A trial with various
samples of zinc oxide showed that it might be relied
upon as a white which could always be reproduced and
one which could be readily obtained. The zinc white
should be mixed with a very pure white gelatine or
isinglass, which is dissolved in hot water. The gelatine
solution is used very sparingly, only sufficient being
added to cause the oxide to adhere to the card on which
it is coated. On comparing the intensity of the spectrum
colours reflected from ordinary card and the card treated
with the oxide, it was found that with the former there
was a slight deficiency in the blue and violet, and also a
little in the green as compared with the former. A card
or board should be brushed over with a cream of the
oxide and be allowed to dry, when another coat should
be given it, and then be flatted down with a brush
when set. An ordinary white card placed alongside
will appear yellowish. There should not be the slightest
gloss on the oxide ; it should appear quite matt if the
surface be properly prepared. Another good receiving
surface is plaster of paris which has been set on a fine
ground-glass surface. It is such a surface that Mr.
Lovibond uses with his tintometer.

COLOUR PATCH APPARATUS 47

On the whole, we prefer the zinc oxide surface.

When using the colour patch it is essential that a
definite surface only should be illuminated, and we have
found that if the face of one side of a cube be covered
with the prepared white card and behind it black velvet
be hung, we have an Ideal screen on which to receive the
colour or white or both. A three-sided prism of equi-
angular section would perhaps be better, as then there is
no danger of the sides of the cube being in any degree
illuminated, which might be the case when the Screen
surface is not absolutely perpendicular to the light fall-
ing on it.

The annexed figure shows the arrangement in use.
A, B, C, D, E, F are
made by dovetailing
two boards at right
angles toone another.
These are covered
with black velvet. '
A scale, K, which is
used for measuring
the width of the slits
in the spectrum, is
fixed as shown. The '
cube H is mounted

on a stand such as are fio. le.

found in all chemical

laboratories, and the iron rod passes through a hole
bored in its centre. The cube can be lowered or raised
by unloosing the screw I, and any surface can be pre-
sented to the light by twisting it rouud on the rod.
There is room on the board C, D, E, F for a rod to be
placed for casting the necessary shadows on the face of
the cube. In some cases a surface of flat card (coated

48 RESEARCHES IN COLOUR VISION

with the oxide) has to be employed. A square of the
requisite size is cut out in matt-black paper and fixed
over it by drawing pins. This plan is not so satisfactory
as that described, as the black paper is always to some
extent illuminated, and as it is in juxtaposition to the
white surface it is sometimes puzzling. When the velvet
background is used, it receives the light, but is very little
illuminated, and any small illumination there may be is
not viewed on the same plane as the white or coloured
patch. We have given these minute details, for exacti-
tude in colour measures very largely depends on attention
to such minutiae.

Other white surfaces can be made by pressing mag-
nesium carbonate in an hydraulic press so as to give a
flat disc, which can be cut into any desired shape. The
surface does not appear to be quite so matt as that of the
zinc white.

Scaling the Spectrum.

The method of ** scaling " the spectrum is as follows.
As is well known, metals can be vaporised by the
arc and show " bright-line " spectra. Thus the vapour
of lithium shows a good many lines when its spectrum
is examined on the screen. There are, however, two
specially bright, one in the red and the other in the
blue of the spectrum. Through the aperture of the
slit which is being used for forming the patches these
lines are successively caused to pass and their centres
made to coincide with that of the slit aperture. The
scale numbers for these lines are noted. If sodium and
magnesium are also volatilised, other lines of known
wave-length can be passed through the slits and the
scale numbers read as before. This enables the scale
numbers of the diflferent Fraunhofer lines to be calcu-

COLOUK PATCH APPARATUS

49

lated, and the spectrum will then be " scaled," and the
colours passing through the slit for any scale number
will be known. The following is a table of wave-
lengths for the different Fraunhofer lines and for the
lithium lines: —

Fraimhofer

1

Colour in
Bright Spectrum.

A

Scale Number

and
Bright Lines.

Wave-
Length.

adopted in
. the book.

A, • • •

Dark red

7594

1

« • •

B .

1

G867

61-3

Li .

Red

670.-)

59-8

C .

Scarlet

6562

58 1 1

Na.
D .

Orange }
Yellow ;

5892

50-6

E .

Green

5269

39-«

bM,'

))

\ 5183

37-7

F .

Blue-green

1 4861

30-05

Li .

Blue

4603

22-8

G .

Violet

4307

11-2

U .

Dark Violet

1

3968

\

1

5 "5

The visible spectrum is divided by these lines fairly
equally along its total length. The difference in these
scale numbers by no means corresponds with the
difference in wave-length. If it be required to know the
wave-length of any scale number, it can be ascertained
with great accuracy by calculation. The squares of

the reciprocals of wave-length l—^j will lie very closely

in a straight line if the scale numbers are used as
abscissae. [It is useful to have a chart made on a large
scale and to read off the wave-lengths from the curve.]

The Production of Images in Monochromatic Light,

The colour patch apparatus has a further use,^ which
arises from the fact that on the screen we have a patch

* See Paper No. 18.

D

50 RESEARCHES IN COLOUR VISION

of white light which is the image of the first surface of
the first prism. If, then, we can by any means form an
image of an object on the first surface of the prism, and
then pass a slit along the spectrum, we shall have its
image on the screen in the same monochromatic light
as that which is issuing from the slit. For the purpose
in view, a single large prism is substituted for the two
prisms shown in Fig. 1 1 , and a long collimator with lens
of a sufficient diameter to fill the prism.

Fjg. 18. Fig. 19.

ll 1 1 L

Fig. 17.

The accompanying figures will indicate the arrange-
ment. To show, for instance, the poles of the electric
light in monochrome on the screen, they were so
placed that a beam of light passed through the slit S of
the collimator on to the centre of the coUimating lens
Lg (Fig. 17). A convex lens Lj, of nearly the same focus
as Lg, was placed in the path of the rays, and so adjusted
that a real image of the poles was formed on L3. These
passed through the lens Lg as nearly parallel rays, and as
such fell upon the prism, and then passed through the
remainder of the apparatus as sketched in Fig. 18, where
M is the prism. Lg is a lens to bring the rays to a focus
as a spectrum on ah after passing through a camera, A.
L4 is a lens, shown in the figure as connected with a

• B«^

COLOUE PATCH APPARATUS 51

camera, B, which brings the image of the prism and the
bright image cast on it to a focus at P. By placing
a slit, Sg, in the spectrum, the image cast on P will be in
monochromatic light (that coming through the slit). Lj
should be of such a focal length that it should be as near
the slit as possible. With this arrangement it is very
curious to watch the variations in the brightness of the
arc and of the flame which accompanies the movement
of the slit through the spectrum, and as each variation
can be photographed on a polychromatic photographic
plate, we can obtain records of all that is occurring
(see Fig. 21). Further, by placing strips (Fig. 19) of
spectacle lenses (cut at suitable distances from their
centres) in front of other slits in the spectrum, images
of various colours can be made to faU on P (Fig. 18).
Incidentally, it may be mentioned that investigations
as to the cause of the variable nature of different
flames can be carried out by this plan.

To obtain an image of the sun in monochrome, a long
collimator appears to be a necessity, but the aperture
need not be large. Suppose we determine to have an
image of the sun on P (Fig. 18) of 2 in. diameter, the
image on M need not be more than 1 in. at most. For
this purpose we must have a collimator 10 ft. long.
Two lenses of this focal length can be fixed one at each
end, and a slit in front of that lens which is presented
to the sun's rays. The arrangements followed will be
the same as those given for the electric light. There
appears no difficulty in producing a monochromatic
image of almost any size if the collimator be sufficiently
long and the face of the prism sufficiently large to take
in the whole of the image cast on it.^

1 It should be mentioned that to minimise diffraction the slits should
be used fairly wide. Hence a long collimator such as described and a good

52 KESEARCHES IN COLOUR VISION

The image of microscopic objects can be thrown on
the screen if these objects are well illuminated, and
although dim, yet they can be viewed on the trans-
parent screen P with ease. The images are such that
they can be well photographed.

dispersion will be necessary to obtain the best definition of the sun's

image.

The prism can be replaced by flat diffraction gratings with most satis-
factory results. The gratings employed by the writer had about 6000
and 12,000 lines to the inch. The images were sharply defined, but, of
course, weaker than when the prism was employed. For solar work this
should not be an objection, since there is plenty of light to work with.

CHAPTER V

THE SOURCE OF LIGHT TO USE WITH THE

APPARATUS

We must next consider what should be the source of

light to be used with the two forms of colour patch

apparatus just described. It is evident that the source

must be an intense one when the spectrum is even but

3 in. long, for it has to be remembered that a narrow

slice of light has to be taken from the spectrum, and

that this has to be spread out into a square patch of

light of some 2 in. side. Suppose the width of the

slice of light be ^ .of an inch, and its length 1 in.

Then the area of the beam at the issuing slit is "05

sq. in. The patch of light of 2 in. side is therefore

'05

— = '0125 less bright than the slice of spectrum colour.

The brightness of the spectrum of any source, such as
a candle or incandescent light, is small, and if this were
used the brightness of the patch of light would be so
enfeebled that the colours might be bleached to some
large extent in consequence of its enfeeblement (vide
pp. 97 et seq.).

Fiu'ther, there is but very small intensity in the blue
end of the spectrum, which, even with a strong and
whiter source of light, is only just sufficient to be useful
for measuring purposes. These two facts prevent either
of these sources from being as a rule employed ; hence
we have to cast about to see what light will be most

53

ft.

54 RESEAECHES IN COLOUR VISION

suitable — that is, be readily available, and remain of the
same quality.

One naturally turns to the sun as a source ; but here
again we are met by difficulties, even supposing that
sunlight was always available.

Sunlight,

It will be advisable to enter into some detail as to
the objections to its employment, which incidentally will
also apply to sky light. The light from the sun at mid-
day, even if vertical over our heads, has to traverse the
thickness of the atmosphere before it reaches our eyes.
Except in the tropics, the sun is never vertically over
us, but is at midday at some less altitude, and con-
sequently has to traverse a greater thickness than one
atmosphere. It may be objected that the atmosphere
varies in density, as the greater the distance from the
earth's surface the less is the density. As a matter of
fact, this does not affect the question, except as regards
refraction, and the whole of the atmosphere may be
considered as homogeneous throughout in calculating
atmospheric thickness. The height of this homogeneous
atmosphere is determined by the height of the mercury
barometer. The specific gravity of mercury is 13"6
times that of water, and water 815 times that of air.
As about 30 in. of mercury balances the pressure of the
air, it has been calculated that the atmosphere extends
upwards about 50 miles. As the sun sinks towards the
horizon, the thickness of atmosphere through which the
light passes gets greater and greater, until, according to
Bougier and Forbes, at the horizon it has to pass through
about 35 J atmospheres. (This limit is due to refraction.
For all ordinary altitudes of the sun the thickness is
given by sec 0, where Q is the altitude.) If the air were

THE SOURCE OF LIGHT TO USE 55

totally transparent, the amount of light reaching some
place on the earth's surface would be the same at what-
ever altitude above the horizon the sun might happen
to be ; but there is some small loss of light due to the
absorption by the atmosphere, which may be supposed
to be feebly coloured, and a much larger one due to
the fact that there are innumerable very fine particles
suspended in it, which produce an effect which utterly
differs from those produced by the colour of a trans-
parent body.

Fine Particles in the Atmosphere.

Lord Rayleigh, in a mathematical investigation into
the effect produced by very fine particles in the path
of a ray of white light, found that they scattered the
light in all directions, and that the amount of scattering
depended on the 4th power of the wave-length.

Thus with waves of light with lengths varying as
2 to 1 , sixteen times more of the first than of the second
would pass through an atmosphere charged with small
particles. The greater the number of particles — that is,
the thicker the atmosphere through which the light has
to pass — the greater will be the loss of intensity of the
rays of short wave-length. In other words, as the sun
sinks to the horizon the light which reaches the eye
becomes yellower, until at the horizon it becomes red.

[A pretty experiment can be performed to illustrate
the change in colour which takes place by the passage of
a beam from the crater of the arc light, when a number
of fine particles through which it passes is increased.
Using an optical lantern illuminated by an arc light, we
can throw an image of a small circular aperture cut out
in an opaque plate on the screen, whicli we may suppose
to be an image of the sun. If in the path of the beam

56 RESEARCHES IN COLOUR VISION

we place a flat cell containing a solution of hyposulphite
of soda (1 of salt to 10 of water), the disc still remains
uncoloured, but if we add a small portion of dilute
hydrochloric acid (1 part of acid and 10 of water) to the
contents of the cell, the hyj)osulphite immediately begins
to decompose, and very fine particles of sulphur are pro-
duced in suspension. The image on the screen begins
to get yellow, and gradually becomes orange, and finally
red, the various stages through which the image passes
indicating the diminishing intensity of the colours pro-
duced by the shorter wave-lengths. This can also be
exemplified very beautifully by throwing on a screen
a longish spectrum of the light of the crater of the arc
by means of the lens Lj (Fig. 15), and placing the cell
with fresh hyposulphite solution in front of the slit. The
colour of the light, which is analysed, can be shown by
the patch of reflected light. When the acid solution
is added with much stirring, the first effect on the
spectrum will be a dimming of the violet, then a further
dimming of the same colour, and also of the blue. After
a while the green will, with the colours just named,
begin to fade. The yellow will next follow, and finally
only the red will be light visible. An ocular demonstra-
tion of the loss of colour is very convincing. The colour
of the fine particles does not matter. The particles are
so fine that the light is not transmitted through them
(to any appreciable extent at all events), and wliether
it be small particles of sulphur or of any other material,
such as smoke, the phenomena detailed above will be
observed when a beam of light is passed through them.]
The following table, which has been published,^ gives
the calculated values of sunlight colours after passage
through different atmospheres.

» " Colour Measurement and Mixture," S,P,C.K», and Papers Nos. 8, 9.

THE SOUECE OF LIGHT TO USE 57

Table I.

Light after passing through Atmospheres of the following

Fraun-

Wave-

Thicknesses.

hofer
Line.

Length.
X.

0.

1.

2.

8.

4.

6.
775

6.

7.

8.

•666

32.

A

7594

1-000

•956

•908

•867

•815

736

707

•107

B

6867

1000

•926

•858

795

•735

•684

•632

•583

•542

•086

C

6562

1000

•912

•832

•759

•r.9i

•632

•576

•526

•480

•019

D

5892

1000

•868

•764

•i\66

•5f;9

•494

•428

•372

•323

•001

B

5269

1000

•808

•644

•518

•4-27

•334

•268

•216

•173

■ • «

F

4861

1000

•738

•544

•402

•296

•219

•161

•119

■088

* • •

G

4307

1000

•609

•367

•2-20

•137

•084

•051

•031

•019

- a ■

H

3968

1000

'506

•254

*12S

•071

•03;$

•016

•008

•004

...

This table was derived from a long investigation of
the value of the coeflBcient of scattering due to the
number of particles present, according to Lord Rayleigh's
formula, which may be taken as

where I is the original light before transmission, and
I' that after passage through the particles, X is the
wave-length, and n is a constant. The author found
that the smallest value of n was '0013 when X"* was,
for convenience' sake, multiplied by 10^^, and that the
mean value was '0017. The table is calculated after
using the mean value.

It may be useful to give the approximate bright-
ness of total sunlight when the sun is at various
altitudes : —

With ....

atmosphere, 1*000

At 90° . . . .

1

•840

„ 30» ...

2

atmospheres, '705

„ 19^30' .

3

•694

„ U^'30' .

4

•496

„ ir30' .

5

•417

„ 9°3(y .

6

•303

„ 8^20' .

7

•266

„ rscr .

8

•216

„ 0- . . . .

32

•002

58 RESEARCHES IN COLOUR VISION

The numbers in the third column are derived from
luminosity curve of the sun's brightness, taken by the
method described in Chapter VIII.

The calculated difference in brightness of the sun
is very marked as it approaches the horizon, which
agrees, it is almost needless to say, with observation.

Sunlight at Heights above the Sea.

So far we have only dealt with sunlight at sea
level ; but before going further it is well that we
should note that as we move our place of observa-
tion higher above the sea, the factor n in Lord Rayleigh's
formula gets smaller and smaller as we ascend. During
several years the writer made observations^ of total
sunlight at heights up to 14,000 feet, with the sun
at various altitudes. His plan was to expose to the
perpendicular rays of the sun a standard platinotype
photographic paper for fixed times. Calling to his aid a
fact which he had found, that for visual rays the relative
brightness of sunlight could (except when the sun was
very near the horizon) be measured by taking a single
ray in the yellow (X 5570) of its spectrum, and measuring
the intensity of that ray only, he applied the same plan
to the photographic paper he employed. The platinum
paper would be regarded as a light-registering surface
for all the rays to which it was sensitive, differing,
of course, in amount from the rays to which the eye
was sensitive. He made experiments to find which
single ray in the blue of the spectrum would be equiva-
lent to the total light acting on the platinum paper.
This was found to be a wave-length (X 4240). The
darkening of the developed platinum paper, after ex-

1 See Paper No. 9.

THE SOURCE OF LIGHT TO USE 59

posure for fixed times at different stations, was carefully
measured. The observations made at the widely varying
altitudes were finally calculated as if the variation
was due to the variation of the wave-length (X 5570).
This enabled the factor for the scattering of light to
be found, which would be applicable to every ray of
the spectrum. The observations made during the three
years show that the factor n in Lord Rayleigh's formula
varies as the height of the barometer at the place
of observation. Thus if the n is •0013 at 30 in.
of barometric pressure, it is only '00065 at 15 in.,
and at 10 in. would only be -00043. Enough
regarding sunlight has now been said to show that
it is untrustworthy as a standard ; that even in a
cloudless sky its quality {i.e. the relative brightnesses
of the diflferent rays) varies, and that the variations
differ according to the altitude at which observations
are made.

Sky Light.

The next natural source of light is the sky^„^«nd
here we are met with precisely the same kind of
diflficulties which are found with-^Bunlight. The light
which is scattered away from a sunbeam by the fine
particles falls on other neighbouring particles and illumi-
nates them, and part come to the eye. Lord Rayleigh
made an investigation into the light irom the sky and
found that the light coming from the fine particles as
" sky " light was more or less polarised, the polarisation
taking place most- strongly in a direction at right
angles to the direction of the beam of sunlight falling
on the eye, and that its blueness was due to the
greater scattering of the rays at the more refrangible
end of the spectrum. The light coming from the sky

60 RESEARCHES IN COLOUR VISION

and the sunlight reaching our eyes, if mixed together,
might thus give us the original colour of the sun-
light as it issues from the sun itself It is perhaps
impracticable to make such a mixture owing to the
fact that a proportion of the scattered light must go
away into space, but it indicates that sunlight at noon
on a summer s day must be slightly less blue than the
light which enters the atmosphere. The polarisation
of scattered light can be shown in a simple manner,
and the experiment is one which imprints the fact
upon the memory.

Polarisation produced in Scattered Light.

[If we take a cell some 3 or 4 in. long and
pass a thin pencil of light through its length, such,
for instance, as is given by sending a beam of light
through a small circular aperture placed in an optical
lantern, there will be no appearance of the light in
the interior of the cell. If, however, we fill the cell
with water in which common mastic varnish has been
precipitated, the turbid liquid at once shows the track
of the light and becomes illuminated. The pencil of
light will appear whitish at the end of the cell near
the aperture, and will be seen as yellower when it
approaches the other end. (A screen placed near this
end of the cell will show the colour of the pencil after it
emerges.) If between the lantern and the cell is placed
a NicoFs prism which is rotated in one direction, the
track of the pencil, when observed at right angles to
the direction of the pencil, will gradually fade away,
and will finally become invisible, as will the illumina-
tion of the water ; whilst if it be further rotated 90^,
the track and the water in the cell become visible once

THE SOURCE OF LIGHT TO USE 61

more. In this experiment the small particles act like
the small particles in the air.'

This investigation of Lord Rayleigh's, which General
Festing and the writer, it is believed, were the first
to confirm experimentally, enabled an experiment to
be made first of all by Sir George Stokes, by which
the debated point as to whether a candle or gas flame
was luminous owing to solid particles being rendered
incandescent could be settled. If the pencil of light
(sunlight by preference) be directed through a candle
or gas flame instead of through the turbid medium, a
track of the pencil can be seen when examined at
right angles to the pencil. When the Nicol's prism is
inserted and turned in one direc-
tion, the track will be invisible ; if
turned in the other direction it will
reappear.

Fig. 20 gives copies of photo-
graphs made of the phenomena.

Such evidence tends to prove
that the particles are solid, though

FiO. 20.

extremely fine. In other words,
there does not seem to be much difference in the source
of light from an incandescent electric light and that of
a candle flame : both appear to be due to incandescent
solid carbon. It may, however, be remarked that the
illumination given by a flat gas flame, when it is turned
flat side towards an object, will not be quite the same
as that given when the flame is turned end on. The
reason for this is apparent.]

* It may be stated that the Bnspended particles become finer if the
water be allowed to rest for a month or two.

62 RESEARCHES IN COLOUR VISION

Nature of Atmospheric Fine Particles.

It has been a somewhat disputed point as to what
the fine particles in our atmosphere consist. Lord
Rayleigh, in a more recent paper than that referred to,
has calculated that the sizes of the molecules of the
gases which make up the atmosphere are sufficiently
large to cause the sky to be bluish, but they can hardly
account for the deep blue which is often seen overhead.
It seems more probable that the main sources from which
the blueness is derived are dust and the particles of water
which are in a semi- vaporised condition. These would
amply account for it. We have often good circum-
stantial evidence before us that such water particles will
produce the effect required. The sky is not only above
us, but it is everywhere above the ground. We often
look at distant hills and find that they have a blue
haze in front of them which profoundly alters the local
colouring. Or again, if we look at a very distant snow
mountain we find that not seldom the whiteness of its
snow is tinged with a yellow which can only be due to
the passage of the white light reflected from it through
fine particles which intervene between the eye and the
mountain. There are dry days when this is seen to the
greatest advantage. When the atmosphere is moist, it
is a matter of common observation that distant hills
show their local colour, and stand out so that one can
" almost touch them." On these same kind of days the
snow of the far-distant snow mountain will appear white
and not yellow. On such a day we have the fine
particles coalescing from bigger drops or particles which
are too coarse to scatter the light, and hence no large
amount of blue is produced by scattering. From obser-
vations and calculations made, it almost appears that

THE SOURCE OF LIGHT TO USE 63

aqueous particles are of two sizes, one of which is quite
small enough to be compared with a wave-length of
light, which is a measure of the suitability of the
particles to scatter light, and the other considerably
larger, and does not scatter selectively. Be this as it
may, evidence goes far to prove that aqueous particles
can give rise to the phenomena of " scattering."

Even were the sky free from cloud, it is unsuitable
for the purpose of a source of light, for the greatest
intensity available is only a disc, which has the same
angular dimensions as the coUimating lens when viewed
from the slit.

Light from the Crater of the Positive Pole,

As already indicated, the best and most constant
source of light to obtain a measurable patch is the crater
of the positive pole of the electric arc light, and this
involves the use of a direct current of electricity. The
crater is a small circular to oval space on the positive
carbon which is at an intense white heat, and if a
" cored " carbon is used for the positive pole it appears
as an almost uniform surface, probably in a semi-liquid
state. The violet rays of the arc are present, but if the
negative pole be the top pole and be placed a little in
front of the positive pole the spectrum of these rays
is reduced in intensity and practically does not interfere
with the far stronger spectrum of the white-hot crater.
In Fig. 11a lens is shown in front of the collimator slit.
This is so placed that an enlarged image of the crater
is thrown on to the slit, filling it completely, and if the
diameter of the lens is sufficient the coUimating lens
will also be entirely filled. Fig. 21 shows six different
images of the poles of the arc light taken in — (1) red ;

64 EESEAECHES IN COLOUR VISION

(2) orange; (3) yellow; (4) violet; (5) blue; and (6)
green monochromatic light (see p. 50). In the red
image the photograph shows the positive pole luminous
— that is, red hot — some distance from the crater. In
the orange image the heating apparently does not

extend so far. The yellow, blue, and green images
show less of the positive pole, i.e. shorter lengths of
the carbon are luminous. The violet image shows
only the crater as heated to " violet " heat. Here
we have evidence that the temperature of the carbon
at different distances from the crater varies. Some
distance below the points we have red heat, then

THE SOURCE OF LIGHT TO USE 65

it has yellow heat ; finally an intense white heat is
generated in the crater. This white heat is practically
constant and of uniform temperature. In fact, the
photograph taken with the red rays shows where the
carbon becomes red hot, and with the green rays where
it has a temperature intermediate between that of the
crater and red heat. It may be remarked that the size
of the crater varies with the size of the carbons and with
the current employed. In our own sloping lamp the
carbons are 13 mm. in diameter, and the voltage 115
volts, and about 11 amperes of current are used. The
diameters of the oval crater are about 7 '5 and 5 mm.,
which are enlarged by the lens from two to three times.

Arc Lamps.

As regards the light, it is advisable, for the sake of
comfort, to use an automatic lamp with the positive pole
remaining always at the same height. The sloping lamp
we have used is a Brocky-Pell or else an Oliver lamp
(by preference the last). Where there is an assistant
to attend to the lamp, one of the comparatively cheap
" scissors " motion lamps can be used, and is satisfactory,
the image of the crater being kept on the slit by the
movement of the ** scissors." The light may be placed in
a darkened room in a lantern which practically cuts off
all light except that coming through the lens used to
give the image of the crater on the slit. It is con-
venient to have the lantern outside the darkened room
and to admit the light through an aperture made in the
wall. This leaves the darkness in the room practically
complete, and for some purposes this is necessary.

The quality of the " crater " light with these two
lamps and with the same carbons never seems to vary —

E

66 RESEARCHES IN COLOUR VISION

that is, the relative brightness of the different rays do not
alter, though the quantity of light forming the spectrum
may diminish to some extent if the slit has not been kept
entirely covered by the crater's image. For this reason
the device of using the reflected beam as a comparison
light is of the greatest use. More recently the writer
has been using a lamp with the carbon for the positive
pole in a horizontal position, the negative carbon is
below and nearly at right angles to the other. The
carbons are larger and take about 22 amperes at 100
volts. When this light is employed, its "quality" is
a little different to that just described, the spectrum
increasing in brightness in the yellow, green, blue, and
violet. This may be due to the greater amount of light
from the very hotest parts of the crater falling on the
slit, or to some necessary alterations that were made in
the optical arrangement outside the slit. When this
lamp is used, the relative luminosities of the different
rays remain the same.

Nevnst Lamp.

Quite recently the Nernst lamp has been used in the
writer's laboratory as a source of illumination. The
means by which it becomes workable was devised by
Professor W. Watson. The Nernst lamp is on the same
principle as a glow-lamp, but in some forms the filament *
is single and of such a length that the whole of it can
be placed in the collimator tube. Professor Watson
employed the white-hot filament in place of the slit and
at the focus of the coUimating lens. The diameter of
the filament is so thin that it answers for a slit of
fairly open aperture. The white-hot filament is enclosed

* It is not a carbon filament, but is composed of a compound of rare
earths such as cerium.

THE SOUBCE OF LIGHT TO USE 67

in a metallic box, which is practically light proof and
which can be removed from the collimator when re-
quired. The spectrum of this light when the current
passing is 1 ampere and the voltage 100 is bright.
By using a ** combining lens" for the spectrum of
shorter focus a smaller patch is formed on the screen
sufficiently bright to be readily measured. Such a light
has the advantage of being perfectly steady. It is too
early to state that the quality of the light remains the
same, but measures seem to point to the fact that the
light emitted from different filaments is always of the
same quality as long as the amperes and voltage are
maintained constant.

CHAPTER VI

THE APPARATUS TO ALTER THE INTENSITY OF

THE LIGHT

It is necessary to have some means by which the
intensity of the light coming through the spectrum
slits, or that of the reflected beam, can be altered
at pleasure whilst observations are being made, and
the writer has found that the plan of rotating sectors
in the beam, with a good velocity, will give results
which compare favourably with that of moving a
comparison light.

Sector Apparatus.

The figure shows a sector which can be opened and
closed during rotation in a very simple way. One
sector (the sector S) is attached to an axle, M, and
the other sector (SO is attached to a hollow axle, N,
fitting accurately the axle M ; a sleeve, A, fits over N.
In the axle M a spiral channel is cut, in which a pin
with a rounded head, fixed to the sleeve A, runs.
A lever, fixed to a support (not seen in figure), carries
a fitting which clasps each side of a projecting boss, B.
When the lever is pushed to the right or left, the
boss moves with it, and at the same time the pin
attached to it travels in the spiral channel on the
axis and compels the sector S' to open or close the
apertures between the segments. A pulley, C, is
attached to the axle, and a leather or thread band
passes over it and the pulley D, which is attached to

APPARATUS TO ALTER INTENSITY 69

a motor, E. Wtien the sectors are rotated by means
of the motor, the apertures can be opened or closed
by the lever H at will. The rims of the sectors are
graduated in degrees of arc.

There have been various attempts made at some
time or another to prove that the sectors do
not give a diminution in light proportional to the

degrees of aperture. It was not till after exhaustive
trials that the writer adopted the sector method for
the purpose of assisting in colour measurement. Lights
of various colours were reduced by the sectors against
a light which could be moved away from a screen to
any required distance, and in no single case did the
sector give any other value than the correct one. It
should be stated that the sectors are only applicable
for accurate measurement when the angles of aperture
lie between 180° and 10°. There is always a small

70 RESEARCHES IN COLOUR VISION

amount of backlash with the sectors themselves,
which, when the angles are smaller than about 10°,
might cause an appreciable error in the measures. The
error is so small in good instruments, that when bigger
angles are used it becomes trifling in comparison with
the errors which may be expected in all such visual
observations. (If anyone wishes to make sectors of
this kind, a less complicated plan is to use an
American drill for the axle.)

Annulus Apparatus.

Another plan for reducing light is by what the
writer has called an annulus, which is a gelatine wedge
in annular form. The late Mr. Leon Warnerke brought
out a " sensitometer " (an instrument for measuring the
sensitiveness to light of a photographic plate), in which
the apparatus for reducing the intensity of light
admitted to a sensitive surface consisted of an annulus
of gelatine of gradually increasing thickness, coloured
either by a dye or by incorporating with the gelatine
a powder of any colour which might be desired. Mr.
Warnerke made a mould as follows : In a perfectly
flat disc of steel a circular groove of uniformly increas-
ing depth is cut out by a proper machine till the
ends of the groove form a circle. The depth of the
groove, when tested, was found to increase pro-
portionally to the arc of the circle, and replicas of
the disc, with its groove, are made in non-oxidisable
metal. For our purpose the finest ivory black is mixed
with a semi-liquid gelatine, and when thoroughly in-
corporated the viscous material is poured into the groove,
the top surface of the disc being accurately levelled.
A sheet of worked glass is then laid on the surface

APPARATUS TO ALTER INTENSITY 71

of the disc, and any excess of gelatine is squeezed
out, except a very fine film, which appears colourless.
When the gelatine has properly " set," the glass
plate is removed with the relief of the groove attached
to it. The gelatine annulus is allowed to dry, and is
then ready for use. The writer had a large batch
of these gelatine annuluses prepared, some giving
small differences in the light, which passed through
the thin end and the thicker end of the relief (It
may be said that the relief is so small that no
prismatic effect can be traced.) Others gave a medium
range of increasing density, and yet others a very
steep gradation, quite useful for extinction purposes.
The annulus was tested as to the transmission of
coloured light, and it was found that from the extreme
red to about the G line in the violet of the spectrum,
every ray was equally obstructed. The graduation of
the annulus should give intensities which varied as the
log of the arc. The various annuluses were tested, and
about one out of every three gave a graduation which
was practically perfect.

The following is the method of mounting the
annulus. A hole is pierced exactly at the centre of
the circular disc (as shown in F). The disc of glass,
A, is also pierced with a hole in its centre, the hole
being just of the size sufficient to allow a pin, with
a screw thread springing from a brass plate attached
to the wooden slide, to penetrate. The disc of
glass, F, is pressed on to the pin, and the two glass
plates are clamped together by a mill-headed nut, D,
a washer of paper, E, being placed between the two.
The disc, A, is cemented into a circular ring, B,
graduated into degrees. On A is ruled a line joining
the centre and the zero of graduation. The junction

72 RESEARCHES IN COLOUR VISION

of the most opaque and transparent parts of the annulus

is made to coincide with the zero point and the line
ruled on A. In the wooden slide is placed a metal

APPARATUS TO ALTER INTENSITY 73

slit, S, with movable jaws opening centrally. When
vertical, the line ruled on A passes through the centre
of S. In the wooden slide G a transparent scale similar
to that shown in Fig. 12 is inserted. The brass circle,
C, can be caused to move round its centre by a thread
passing over it and a small-toothed pulley, to which is
attached a long arm, B, that causes the pulley to rotate
when it is turned.

[The annulus in ordinary use has regular gradation
for each degree, the coefficient of obstruction (it is not
exactly absorption) being 0*0086 for each degree.]

To use the annulus in the spectrum, the slide
bearing it is placed in the place of the slide D of
the colour patch apparatus. When using it in the
reflected beam, the slit S is placed in the position
where the rays from the reflected beam cross, and
which is really the image of the collimator slit. By
the long arm mentioned, the annulus can be rotated
and the intensity increased or diminished. A com-
parison of measures between the sector and the annulus
shows the results to be identical.

CHAPTER VII

INTENSITY OF SPECTRUM COLOURS

The first and simplest measure of colour to make is that
of the intensity of the spectrum colours which are trans-
mitted through, or reflected from, coloured bodies. The
various methods which we have adopted will be de-
scribed in this chapter. The intensity of a spectrum
colour transmitted (or reflected) we will define by the
percentage brightness that it bears to the same colour
unmodified by transmission or reflection. Thus, sup-
posing it is found that after transmission through a
green glass, the sodium light at D is (after making
certain corrections), only half as bright as that which
falls on the glass, then the intensity of this colour is
•5, or 50 per cent. Evidently it is convenient to have
what we will call the naked light compared directly
with that which passes through, or is reflected from, the
medium. The first method that will be described is
the latest in point of date, and is perhaps the most
satisfactory.

Modification in the Apparatus to form Two Beams

of the Same Colour.

The single colour patch apparatus may be used for
the purpose. A single slit is used in the slide at D,
Fig. 13. Between it and the colour patch is placed on
a suitable block a bundle (Mj) of plane and colourless
glass plates (Fig. 11) about 5 in. long by 3 in. deep.
The 5-in. length makes an angle of about 45^, with the

INTENSITY OF SPECTRUM COLOURS 75

ray issuing from the slit S, and the 3-in. side is vertical
(Fig. 24). The bundle is so placed that the whole of the
spectrum has to pass through it after it has passed
through the lens L,
which forms the

^

§

Sector

patch on the screen.

The glasses and the

bundle are separated

one from another by

the thickness of a

strip of paper at

their edges. Any ray falling on the bundle is divided

into two parts; one is reflected about 90° from its

original path and the other passes through it as shown

in the same diagram. The silvered mirror Mg reflects

the light from the glass bundle on to the screen, and

forms a patch which can be superimposed on the patch

formed by the direct beam.

A sector can be placed in the path of either beam, so
as to diminish their light at will.

The amount of light which should be reflected from
the bundle, it is often supposed, can be calculated from
the number of individual plates in it, when the angle of
incidence at which the light falls is known.

The calculation is, however, not always to be relied
upon, owing to defects in the glass, want of perfect paral-
lelism of the surfaces, and the variable absorption. It is
easier to determine experimentally the total amount that
is reflected. The following table gives the results of some
measures made with the bundle we employed when the
incident beam made an angle of 45° with the surface : —

1 glass reflects 12*5 per cent.

2 glasses reflect 22

3 ,, «, 28

»»

4 glasses reflect 32 per cent.

5 „ „ 34

6 „ „ 35

76 RESEARCHES IN COLOUR VISION

It will be seen that after six glasses are in position,
there can be no very little marked alteration in the
percentage reflected. Of course, the amount reflected
has to be deducted from the total amount coming
directly on to the screen, besides that which is lost
from absorption by the glasses, which, it may be stated,
is by no means small.

Measurement of Absorption of Transparent Media

and of Pigments.

The transparent medium the absorption of which
has to be measured is placed in the direct beam of
light. Two shadows, side by side and touching one
another, are cast on the screen by placing a rod in
the path of the two sets of rays. One is illuminated
by the direct ray which comes through the medium
whose absorption is to be measured, and the other by
the reflected beam which has not passed through it.
The illumination of the two shadows are equalised by
placing the sectors in the path of the reflected beam.
If necessary, another set of sectors, set at known angles,
can be placed in the other beam. The writer usually
commences with a ray in the red. The percentage of
loss of the direct ray owing to absorption is ascer-
tained by substituting for the transparent medium
a colourless glass and again equalising the shadow
illumination.

Let us take as an example a green glass, the ab-
sorption of the D (sodium) light by it being required.
With this glass in position, the rotating sectors in the
reflected beam showed 15° of aperture as necessary to
equalise the shadow illumination, but with the colourless
glass it required 36°, the sectors being in the direct

INTENSITY OF SPECTRUM COLOURS 11

beam in both cases. The percentage intensity of the
ray passing through the medium was therefore 41*7 per
cent, of the original beam.

The simplest way of calculating the result when the
sectors have to be changed from one beam to the other
is to multiply the readings by one another and by 100,
and to divide by (180°)^ In the above example, if
the sectors had been placed in the direct beam for the
second reading, we should have —

L5x^36j^l00_ 100_

n^oxTso" " 60

— that is, the percentage of light transmitted would be
1*67. Taking another case, the reading of the sector in
the direct beam with the colourless glass interposed was
as before, viz. 36°, but in order to equalise the shadows
when the coloured glass was in position a second sector
had to be inserted in the path of the direct beam, which
was fixed at 60°, and the reading of the moving sector in
the reflected beam was 62. Had the direct beam been
left without a sector, it is evident that the reading
would have been 186°, since only one-third of the direct
beam was allowed to pass. As before, the actual read-
ISO
ings are multiplied together, as also by 1 00 x -^ and

180
62x36x-^^xl00

divided by 180^ that is, ^,^ ^^^^ = 207 %.

^ ' ' 180x180 ^°

These calculations are of course done after the obser-
vation. It may be said that at least three readings
should be taken for each scale number, and the mean
used for the calculations.

The following is a complete table of the observations
and calculations made for ascertaining the intensity of
light passing through an inch of a saturated solution of

78 RESEARCHES IN COLOUR VISION

Tablb II. — Intensity of Liyht trantmxUed Ihnntgk PoloMium
C/iroiitale ; alio the Lumitumti/.

Tmiismitted

(Horiiontol

Light

Carbon.)

N.ked Light
bomg 100).

Light.

68

1

-68

75

2

1-5

8.3

8-7

7-2

88-5

21-3

18-8

91-5

48-3

44-2

92-5

70

64-7

94

84-7

796

95-5

96-2

91-9

97

100

97

99

95

94

100

85-3

85-3

93-5

72

67-3

77

561

432

00

41

20-5

16-5

27-5

4-5

2

I5'8

32

INTENSITY OF SPECTRUM COLOURS 79

chronfate of potash. The luminosity is shown, as it will
be required to be known later.

When the intensity of the colours reflected from
pigments Is required, very much the same procedure
is followed. The only difference is that one half of a
square surface of, say, Ij-in. side is covered with the
pigment and the other half with white, the shadow
illuminated by the direct beam falls on tlie pigment,

and the white is illuminated by the reflected beam. The
equalisation of the illumination of the shadows is effected
in precisely the same way as that described when the
colour intensities of the transparent medium were being
measured. The percentage reflection is arrived at by
substituting a square of white paper for that made up
of pigmented paper and white. The rod, of course, is
employed to cast the two shadows in each case. The
diagram (Fig. 26) and the following table show the
light reflected from a specimen of emerald green.

EilSEAfi'.HSs^ IN •:»:»D>rR VISION

fr — -.• "

II — L.tit*rft-

Mff^.

m. -*» ^ ■■

Orcimt^il

*r^^.'~rwi.

a

LnhrtT^

Oriiuoe

?*Uh^ >

UraT^

fr«

3""3

•m

4

*5

Daurraai-

'.A

^».—

^♦3

3-5

«r-

^ :

4

3 5

35

^.

4

*'T

3 5

3-5

>•

4

4 '

4

4

4

5i?

»

-f

5

5

5

^

•

^

'^ ~

^

«s

5^

- -*

'1 *

.4

14

14

-:••

^M ■

^

^*

±^

27-5

4^

4i

4:-'^

4±

-tr

41-5

^

r.1

^^

'^i

54

55

^i

"^3

•^-

•r±-5

«

63

A±

• m.

71

71

¥t

74

74

74

74

74

»

74

7'»

74 '>

745

74-5

»

71

73

73

73

73

^

70

7*>

*^3

70

70

32

►V,

**4-.'.

•)«

65

66

3r#

61

6:-.-»

«i

61

62

^

%^

-"•7

59

5S

57

26

52

.>4

53

53

52

24

46

46

46

46

46 ,

i2

40

4»i

40

40

-W

20

Z\

32

31

32

34

18

27

27

27

27

27

16

22

ei-5

22-5

22

22

14

17

16-5

175

17

16

12

10

12

14

12

10 !

10

5

3

5

5

6

8

3-5

3-5

3-5

3-5

, 3-5

6

3-5

• a •

••■

3-5

' 35 .

4

3-5

• * «

• • •

3-5

3-5 i

1

1

Light reflected from white =100.

Alternative Method of Measurement,

Another plan of measurement which is suitable for
the colour patch apparatus is to place a double image
prism against the lens of the collimator. This will cause
two similar spectra to be formed, one above the other.
The separation given by the prism should be suflBcient
to leave a blank space some quarter of an inch wide

INTENSITY OF SPECTKUM COLOURS 81

between the two spectra. A long slit passing through
both spectra takes the place of the shorter slit usually-
employed in the spectrum. The double image prism is
turned so that the same colour comes through the slit
from the two spectra. In front of the top part of the
slit a right-angled prism, A, is attached to the slide
carrying the slit. This reflects the rays coming from
the top spectrum along the slide, and these are again
reflected by a second right-angled prism, B, on to the
screen. The rays from the bottom spectrum go direct
to the screen on to the

same square as that ^ j.

on which the reflected n ' "^

beam falls. A rod ^ \ ^

placed in the path of ^'^- ^^•

the rays causes two shadows to be cast, which are illumi-
nated as before. It is convenient to have the two right-
angled prisms attached to ball and socket joints, which
can be fixed by screws. The ball and socket is attached to
a knitting-needle, which passes through a hole in a brass
plack which is attached to the slide. This enables the
patch of light to be adjusted to fall on any desired part
of the screen or on one side of a cube. The illumination
of the two shadows are equalised as before, and if the
same colour ^ passes through the slit from each spectrum
the sectors should not require to be altered when the
colours from the naked spectra are used ; any alteration
shows that the double image prism requires adjusting.
The light coming through transparent media and re-
flected by pigments may be measured by this apparatus,
the necessary parallax being obtained by the distance of
the second from the first reflecting prism.

^ The D light is the best light to use for adjusting the spectra, as a
minute error can at once be detected by the colour of the two patches.

F

82 RESEARCHES IN COLOUR VISION

Disc Method of Measurement.

Another simple plan for the measurement of the
intensity of the colours reflected from pigments is to use
a revolving disc the outer ring of which is made up of
adjustable black and white discs. The centre is covered
with a disc of paper on which the colour to be measured
has been spread. [Pigments can be readily painted on
white paper of such a thickness that the white of the
paper is completely hidden. A few drops of hot
gelatine solution are dropped into a mortar and the
pigment well mixed with it, a little hot water being
added till it is sufficiently fluid to enable a hog s bristle
brush to take it up. The paper is pinned on to a board
and the brusli worked up and down and across till it

appears evenly coated. It is then

allowed to dry, and a second coat
D given.]

The appearance of the disc is that

shown in the figure.
p^Q 28. ^ ^^ white paper wliich has been

coated with zinc white. B is black
paper which has been coated with ivory black ; it
should be dull and matt. C is the pigment, w^hich
is also matt. D is the screw which attaches the com-
pound disc to the shaft of the small motor which
rotates it.

The white and black discs have a cut made radiall}'
from the centre, so that they can be interlocked as
shown (see Fig. 45). There is a small amount of white
light reflected from the black surface, and this has to
be determined. The most convenient method of making
the determination is by the colour patch apparatus. A
square surface of about 1^ in. is half covered with the

INTENSITY OF SPECTEUM COLOURS 83

black and half with the white. The black should be
illuminated with the recombined spectrum white and
the white surface by the white reflected beam, and a
rod be used to cast two shadows. The illumination of
the shadows is equalised as before, and knowing from
measurement the ratio of the two white lights, the
percentage of white reflected from the black pigment
is calculated. A good black should not reflect more
than 3|- per cent, of light, and should be the same for
every colour.

To ascertain the amount of light reflected by the pig-
ment, the compound disc is placed in the colour patch
as shown. The outer ring is given a
known proportion of black to white.
The disc is rotated and the slit through
which colour issues is moved along the
spectrum until a place is reached where
the central disc and the outer ring
both appear to be equally dark. The

- ""^ * . /» 1 . A IS the pigmented disc.

scale number of the spectrum colour is b is the wnck and white
read off, and the proportion of black c is t^e'cdour patch.
to white altered. The disc is again
rotated, and a reading obtained as before. It must be
remembered that with certain pigments, such as green,
there are two places in the spectrum where the equality
of illumination between the centre and the ring appears
the same, and in some few cases there may be more
than two places. It should be ascertained before the
measures are finished that there are sufficient scale
numbers noted to enable the results to be shown
graphically without large gaps appearing between the
ordinates. To show the intensity graphically, the
abscisssB are the scale numbers and the ordinates the
percentage of white which is used. This last must

84 RESEARCHES IN COLOUR VISION

take into account the white light reflected from the
black.

To take an example of the calculation, we will
suppose that the black occupies 270° of the circle and
the white 90°, and that the white light reflected is
4 per cent.

Four per cent, of 270 is 10*8, so that the total white
is 90 + 10-8 or 100*8°. The percentage of the colour

reflected from the central disc is therefore - — xlOO

360

or 28 per cent.

[It is convenient to divide the circumference of the
circle into 100 parts, so that the readings are easily
calculated.

In this case the readings would be 75 black and 25
white, 4 per cent, of 75 is 3, and the white used is
25 + 3 = 28 per cent, as before.]

Measurement of Iridescent Colours.

There are instances of reflection which cannot be
dealt with quite so simply. Take, for example, the
colour of glass flashed with silver. By transmitted light
the glass is canary coloured, but by reflected light a
beautiful peacock blue. To obtain the intensity curve
of the blue is somewhat difficult. The piece of yellow
glass is backed with black backing in shellac, so that
practically no light can be reflected from the back
surface. A piece of white paper is pasted on the
flashed surface of the yellow glass, and a black mask is
cut which allows a rectangle of the iridescent surface of
the glass to show and an equal rectangle of white. The
glass is placed at such an angle that with white light
the iridescence is seen. The bundle of glasses, as

INTENSITY OF SPECTRUM COLOURS 85

before, reflects part of the ray, and the light trans-
mitted by the bundle falls on the iridescent surface,
whilst the reflected beam falls on the white surface and
is used for a comparison light. By placing the eye
opposite a hole cut in a card fixed in the proper
position, the surface is always viewed at the angle
which gives the maximum iridescence. The readings
can then be made as before.

CHAPTER VIII

THE MEASUREMENT OF LUMINOSITY

In the last chapter it was shown how the intensity of
the colours of the spectrum transmitted through or
reflected from coloured objects could be compared with
the same colours of the naked spectrum reflected from
a white surface. In this chapter it is proposed to show
how an estimate of the brightness or ** luminosity " of one
colour can be compared with that of another. This
is a totally different problem to that of comparing the
brightness of lights of the same colour. Suppose some-
one is given pieces of red and green pigmented papers,
and is asked how the brightness or luminosity of the
colours reflected from each can be compared, the usual
reply would be that it is impossible to make any com-
parison between them. We shall see, however, that
it is not only possible, but perfectly practicable, to
obtain very close values of their relative luminosities.
It will have to be recollected that in estimating lumi-
nosities, the nature of the light by which the colours are
illuminated has to be stated, as they will vary consider-
ably according to the whiteness of the light in which
they are viewed. The " colour patch apparatus " is one
means of ascertaining the luminosities of such colours as
those named above when the illuminating light is sun-
light or the arc electric light. It must be here stated
that practice is required to make accurate luminosity
measures of two such different colours.

80

THE MEASUREMENT OF LUMINOSITY 87

The Com/panson of Luminosity of two Pigments.

A beginner will find it easier to make a comparison
of a bright colour with a neutral colour, such as white,
rather than with another bright colour. When two
colours are compared with a neutral colour, it is easy to
calculate the relative luminosities of the bright colours.
For instance, let us suppose that by some means it is
ascertained that the brightness or luminosity of the
light reflected from the red pigment is 25 per cent, of
that reflected from the white (a neutral colour), whilst
that from the green is 35 per cent. The ratio of the
luminosities of the two colours is evidently 25 to 35,
or 5 to 7.

In order to make the comparison of the red with the
white, a rectangular piece of pigmented paper, say,
1 in. X ^ in., is placed alongside white paper (the
white being oxide of zinc) of the same
size, and the two are surrounded by a
black mask (Fig. 30).

The patch of white formed by the
recorabined spectrum is thrown on the
coloured paper K, and that from the re-
flected beam on to the white rectangle W, the two white
beams being separated by placing a rod in their paths.

[It is sometimes convenient, in order to do away with
fringes which may appear in the combined white owing
to the different rays striking the rod at slightly different
angles, to arrange the recombining lens of the colour patch
apparatus so that the edge of the white image of the
prism falls on the junction of the red and white patches,
and only to use the rod for the purpose of casting a sharp
shadow from the reflected beam on the white surface.]
We will here suppose that a sector with movable

88 RESEARCHES IN COLOUR VISION

angular apertures is placed in the recombined beam.
When the aperture is wide, it will be seen that the red
is evidently brighter than the white. The aperture is
then much reduced, when it will be felt that the red
is darker than the white. Evidently there must be
some aperture of the sectors which will transmit the
exact quantity of light which will make both red and
white of the same brightness. The angles of the sectoi*s
are rapidly altered from " too light " to " too dark "
and back again, and the range of angle is gradually
diminished until the observer sees both to be equally
bright. The angle is noted, and the observation re-
peated, till the readings become concordant. The
mean aperture is taken as the aperture which gives
equal brightness to the two rectangles. Say that the
mean aperture is 48*^. The red rectangle is then re-
placed by a second white rectangle, and the luminosities
equalised, the mean apertiu'e of the observation giving,
say, 12°. The red is therefore a quarter (or 25 per
cent.) of the brightness of the white. The green pig-
mented paper is treated in a similar manner, and the
reading is, say, 34°. This makes the green ^^, or 353
of the white. The ratio of the luminosity of red to
green is therefore 25 to 35'3, or about 5 to 7.

In making these measures, as already said, at least
three readings should be taken, and in difficult matches
even more should be made. However, if the mind has
been fixed on the necesi^ity of noting the " too light "
and " too dark " oscillations, the mean of three readings
should be sufficient in most cases. When the rect-
angles are of the size given above, all observations
should be made with the eyes at a fixed distance of
about 5 feet from the patch, so that the images may
all be received on the yellow spot.

THE MEASUREMENT OF LUMINOSITY 89

Luminosity of Pigments in Artificial Light

If the luminosity of the pigments in artificial light is
required, the following plan may be adopted.

L is the light, M a silvered mirror, the pigment is
illuminated by the reflected beam, and the white by

^

-^ „ Sector I

Rod m\ ^ I

Fig. 31.

the directed beam d. A screen SS, with two apertures
cut for the rays to pass through, is placed in the path
of the two beams, and the luminosity determined as
before. The light must be enclosed and observations
made in a darkened room.

It cannot be too strongly impressed upon the
reader that it is absolutely fatal to good results if stray
light is allowed to fall on the white and the pigment, as
there is selective reflection from the latter, which is by
no means the same as that from the white surface.

Comparative Luminosities of Spectrum Colours
as seen on the Yellow Spot.

An exact determination of the comparative lumi-
nosities of the different rays of the spectrum itself is
all-important.

It should be carried out in precisely the same manner
as that just described, but with the colour patch appa-

90 EESEARCHES IN COLOUR VISION

ratus. Instead of a pigmented paper being illuminated,
the whole square is white. On half of the square the
patch formed by the rays coming through a slit, which
can be moved along the spectrum, and on the other
half the reflected beam, falls.* A rod in the path of
the converging beams prevents the overlapping of the
colour and the white, and the two can be caused to
touch by adjusting it. The rotating sectors have usually
to be in the path of the white beam, and the oscilla-
tions of aperture will thus alter the luminosity of the
white.

The slit which passes along the spectrum, of course,
remains unaltered in width during the whole of the
measures, so that the luminosities of the different rays
are strictly comparable one with the other.

When the blue end of the spectrum is approached, it
will be found that the readings of the sector apertures
become very small, and, owing to a small amount of
backlash, which almost of necessity exists in the sector
movements (see p. 69), they may become unreliable. It
is usual to substitute for the silvered mirror, which
reflects the white beam, a piece of flat unsilvered glass.
The ratio of the reflections of the two mirrors are very
readily determined, and the readings of the unsilvered
mirror can be converted into readings of the silvered
mirror when once this has been found. Sometimes it
has been found useful to place in the white beam a piece
of blue glass, which practically absorbs all the rays
except the blue and violet. When the absorption by
such a glass has been found, the readings, as in the case
of the plane mirror, can be converted into readings with

^ Care should be taken that the centre of the colour patch should fall on
the centre of one half square, and the centre of the white patch on the
centre of the other.

THE MEASUREMENT OF LUMINOSITY 91

the silvered mirror. (For rather smaller diminutions of
luminosity, a piece of wire gauze, placed in the path
of the white beam, is effective, the diminution being, as
a rule, rather more than half.) Some observers find
it an advantage to have the white comparison light
thus converted into a blue one, as the colours in the
blue and violet approach that transmitted by the blue
glass. It is again necessary to repeat the warning that
the eyes of the observer must always be at the same
distance from the screen, and that he should be " dark
adapted" {i.e. his eyes should be withdrawn from day-
light for ten minutes before measures are read), when
observations are made, in order to obtain reliable com-
parative readings.

Luminosity of Colours outside Yellow Spot.

For theoretical purposes, it is also advisable to deter-
mine the luminosity of the spectrum when not received
on the yellow spot. To make such observations we can
adopt a plan which, though it appears difficult at first,
is yet easy to carry out after a little practice. In order
that the image of the patches may fall outside the yellow
spot, it should be received on the retina at least 5^ from
the centre of the eye. If a spot is marked in a hori-
zontal direction 5 in. away from the outside of the
rectangles, and the observer s eyes are 5 ft. away from
the patch, and that spot is looked at, the image of the
rectangles will be received outside the extreme edge of
the yellow spot. The outside spot should be illuminated
by Balmain's paint. One eye must be closed, and the
axis of the other eye be directed to that spot. The
rectangles of white and colour will be fairly defined and
the luminosities can be compared. It may appear strange

92 RESEARCHES IN COLOUR VISION

that the luminosities of the two patches can be com-
pared under such circumstances, but as a matter of fact
they can be compared with even greater facility than
when observed with the centre of the eye. When a
comparison is being made, the colour often appears, not
actually to vanish, but to become less powerful (due no
doubt to causes which will be treated of in colour fields),
and to allow matching in luminosity with comparative
ease. The luminosities found appear not to depend on
the azimuth, but to be the same all round the axis
when the spot is moved in a circle round the centre
of the rectangles.

If two square patches of equal size, say of 1^ in.
side, are placed 6 in. apart, and illuminated with white
light of the same intensity, and the centre of the eye be
fixed on one of them, the image of the other will fall
outside the yellow spot. By diminishing the luminosity
of one or the other, the two may be made to appear
equally bright on the two portions of the retina.
Adopting this plan, and taking the mean of a large
number of readings, it was found, to the writer s eyes,
that the relative sensitiveness for white light of the
centre of the retina, and of a spot 10° outside the axis,
was as 37 to 33. The areas of the two curves plotted
from the direct and "10° outside" observations, when
the same white light was employed, were as 167 to 156,
"which is a ratio very close to the above, and thus the
ordinates of each of the curves may be taken to indicate
the relative luminosities of the colours in the different
regions of the spectrum, and they are shown thus in the
tables given below. A reference to Chapter XII. on
the extinction of colour, will show how necessary it is
that in these observations the colour and white patch
should be of equal size.

THE MEASUREMENT OF LUMINOSITY 93

Luminosity of Colours on the Fovea Centralis.

There is another part of the retina on which, if the
different colours fall, the luminosities may vary from
either of the foregoing. The fovea centralis, it may be
remembered, is a very small area lying in the middle
of the "macula lutea," or yellow spot. It is usually
supposed that the axis of the lens cuts the retina in
this spot. In order to arrive at some idea of the lumi-
nosities of the different rays when they fall on this very
small area, a white cube of J-in. edge was employed,
and the colour and white light each occupied one half of
one of the faces. The eye was kept 5 ft. from the small
surface, and the comparisons made in the usual manner,
except that one eye was kept closed.

Calculations from these observations point to the
fovea being about one-sixth more sensitive to the D light
than is the macula lutea. To the green and the blue,
the fovea appears less sensitive than the macula lutea.
If the luminosities be taken at a greater distance than
5 ft. from the eye, it will be found that the fovea is
less sensitive to green, and more to red, than is shown
in Table IV.

This may be verified by causing an image of a star
to fall on the absolute centre of the fovea, and comparing
the colour of an adjacent star with it. The colour of
the two stars will be found to differ, even if in the
telescope they appear the same.

Alternative Method of ascertaining the Luminosity

of the Spectrum Colours.

There is an alternative method of making the lumi-
nosities equal with the spectrum colours. In the white
beam may be placed sectors with fixed apertures, and

94 RESEARCHES IN COLOUR VISION

Tablb Vf.—lMminotity Curvet. {Are light erafer, inelived carbont.)

I.

8ea1e

II.

III.

IV.
Yellow

V.

I.

II.

IIL

IV.

V.

Ware-

Outoide

Scale

Ware-

Outside

Yellnw

Yellow

FoTea

Number.
64

Length.
7217

Yellow

8pot.

• ■ •

Spot.

• • •

CentnUit.

• • •

Number.
32

Length.
4924

1 Cll\#^

Spot.
21

Spot.

8-6

Central ia.

6-5

63

7082

• • •

1

• • •

31

4885

18-5

7

5-5

62

6957

1

2

2

30

4848

16-5

65

4

61

6839

2

4

4

29

4812

14-5

4-7

3-5

60

6728

35

7

8

28

4776

13

4

3

59

6621

7-5

12-5

15-5

27

4742

11-5

36

2

58

6520

12-5

21

24

26

4707

10-5

2-8

24

57

6423

19

33

37-5

25

4675

9-4

2-3

21

56

6330

27-5

50

60

24

4639

8-2

1-82

19

55

6242

35

65

77

23

4608

7-3

1-6

1-5

54

6152

43

80

90

22

4578

6-3

1-4

■ • •

53

6074

52-5

90

97

21

4548

6-7

1-2

■ • •

52

5996

61

96

100

20

4517

5

108

1

51

5219

71

99

100

19

4488

4-5

•94

• « •

50

5850

79

100

98

18

4459

4

•86

• • •

49

5873

84

99

95

17

4437

3-6

•78

« ■ «

48

5720

85

97

90

16

4404

31

•70

• • ■

47

5658

83-5

92-5

85

15

4377

2-7

•62

62

46

5596

81

87

79

14

4349

2-3

•56

• • •

45

5538

77

81

725

13

4323

2-1

•50

• • •

44

5481

72-5

75

66

12

4296

1-9

•45

« • •

43

5427

68

69

59

11

4271

1-65

•40

• • •

42

5373

625

62-5

51

10

4245

1-4

•34

• « •

41

5321

57

57

45

9

4221

1-2

•30

1

40

5270

52

50

40

8

4197

1

•26

• • •

39

5221

46

42-5

32

7

4174

•88

•22

• « ■

38

5172

41-5

36

27-5

6

4151

•75

•18

« ■ •

37

5128

37-5

29-5

22

5

4131

•63

•16

• ■ •

36

5085

33 5

24

18

4

4106

•50

•14

« • •

35

5043

30

18-2

14

34

5002

26-5

14-2

10

33

4963

24

•

10^5

8-4

THE MEASUREMENT OF LUMINOSITY 98

96 RESEARCHES IN COLOUR VISION

the equality determined by moving the slit backwards
towards the red and forwards towards the blue. Some
ray between the maximum luminosity in the yellow and
the extreme ends of the spectrum will be found which
is of equal brightness to the white as diminished by the
rotating sectors. There are, of course, two positions,
one on each side of the yellow, which have equal
luminosity.

The sector may be made of thin card, the alternate
quadrants being cut out as shown in the figure (which

Fig. 33.

is one of a pair), and the rim should be correctly gradu-
ated. By this plan it is feasible to make the double
aperture read 2°, and when one is covered up it will
thus show 1°. These are angles so small as to preclude
them from use with rotating sectors, which open and
close at will during the rotation, owing to the existence
of ** backlash," as has already been said. For still
smaller luminosities, resort must be had to the plane
unsilvered mirror or to the annulus.

The following are the scale numbers, wave-lengths,
and luminosities of the difierent fiduciary Fraunhofer
and some bright lines. The luminosities are those found
with the crater of the positive pole of the electric light,
sloping carbons, withiji the yellow spot on the retina.*

* At pp. 244, 245 will be found the luminosity curves of the spectrum
when formed with the arc light with horizontal positive pole ; also for the
Nernst light and for a paraffin light at page 251 ; Paper No. 4.

THE MEASUREMENT OF LUMINOSITY 97

Table V.

Scale Number.

X

Luminosity.

B . . . .

61-3

6866

1
4

Li (red)

59-8

6705

8 i

C . . .

581

6562

17 1

D . . .

50-6

5892

99-5 1

E

39-8

5269

48

b (Mg.) . .

28

5183

36

F

30-2

4860

6

Li (blue) .

22-8

4603

2

G

111

4307

•6

Luminosity of a Spectrum produced by Feeble Light,

To ascertain the luminosity of a very feeble spec-
trum, a special plan has to be adopted. The compari-
son white beam should be introduced into the measuring
box described at p. 148. In the measures made, and
which are described, the D light when uninterrupted
by the sectors, had a luminosity of 13^ of an amyl lamp ^
at 1 ft. off at the end of the box. The beams from
the spectrum were introduced into the apparatus so
that the colour patch fell on S. The luminosity of the
different rays was taken in the ordinary manner, inter-
posing the rotating sectors in the reference beam. The
following results were obtained (see Table VL),the mean
of the readings being given. Here we have a proof that
the normal eye becomes insensitive to the red end of the
spectrum when formed from a much-reduced intensity of
white light. It must be remarked, however, that all
colour was not entirely absent, though it was very con-
siderably reduced in saturation. The measurements were
made with some trouble at first, owing to the inclination

^ An amyl lamp gives a light closely equal to that of a standard candle.

G

98 RESEARCHES IN COLOUR VISION

of the eye to direct its axis to some point other than the
centre of the patch where the white strip and the colour
strip touch one another. The diversion of the axis of
the eye in some cases made the colour more luminous,
and in other cases less, than it did when the eye was
properly directed.

Table VI.^ — LnminosUy of Sjtect'rum Reduced in Intensify so th<U

^^ = jw5 i4my/ Lamp Iff. distant.

Moan Roadin^;

Scale Nuinl>er.

Ware-leng^.

reduced to
LOO Maximuai.

56

6330

1
5-4

r>4

6152

5

• 1

52

5996

13

50

5850

21 ,

48

5720

42 !

40

5596

66

44

5481

84 '

42

5373

95

40

5270

100

1 38

5172

94

30

5085

84

1 34

500iJ

72

32

4924

58

30

4848

45

28

4776

32

20

4707

23

24

4639

17-5

22

4578

14

20

4517

11

14

4349

5

10

1

4245

2-5

^ These results and those on pp. 100 to 103 are to be found in
Paper No. 4.

THE MEASUREMENT OF LUMINOSITY 99

100 KESEARCHES IN COLOUR VISION

Relative Luminosity of Rays for Different Specti^m

Intensity,

Having found from the curves that the relative
luminosities of the rays of the spectrum when feeble
differed from the same rays when bright, it became a
matter of some importance to ascertain in what manner
the relative luminosities of the rays varied when the
intensity of the light which formed the spectrum was
altered iji a definite ratio. Evidently the most satis-
factory method of ascertaining this was to throw a
patch of white light on the screen and then to diminish
its luminosity by known amounts, and, having selected
some rays of the spectrum, to measure their luminosities.
The box described at p. 148 was again brought into
requisition. A beam of white light was caused to
illuminate one half of the small white square screen at
the end of the box, and the other half was illuminated
by the ray whose luminosity was to be tried. Rotating
sectors were placed in each beam ; the apertures of
those in the white were fixed at different angles, whilst
those of the sectors in the coloured beam were opened
or closed till the luminosities appeared the same to the
eye, a series of readings being taken for each ray. The
results thus obtained were plotted, and some typical
rays are shown in Fig. 35. The ordinates are the
apertures of the sectors which were placed in the path
of the monochromatic rays, and the abscissae the aper-
tures of the sectors in the white beam. The tangent of
the inclination to the vertical of the curve at any point
therefore represents the ratio of the luminosities of the
coloured to those of the white beam for known in-
tensities of light. If this ratio were the same for all
intensities, the curve would become a straight line

THE MEASUBEMENT OF LUMINOSITY 101
starting from the origin. This is only the case, it will

be seen, with one ray, viz. that at scale number 46'3

102 KESEARCHES IN COLOUR VISION

or about X 5618. This ray and white light would there-
fore be extinguished together.

It will be seen, however, from the diagram, that the
other curves become straight lines when certain degrees
of intensity, different in each case, are reached ; and if
these straight lines are produced to cut the axis, the
ordinates of the rays which lie towards the blue end of
the spectrum above 46 "3 have a negative value at the
zero of white light, whilst those which lie toward the
red side of 46*3 have a positive value; showing that
with rays of equal luminosity the blue part of the
spectrum should be extinguished last, and the red
part first ; we shall see in Chapter XII. that this is
the case.

It is, moreover, evident, and this has been demon-
strated by experiments described above, that for low
intensities of light the limiinosity curves of the spectrum
will vary as the intensity is increased, but that a degree
of intensity is soon reached when all the curves in Fig. 35
become straight lines. The distances from the origin
where the lines are curved are so small compared with
the distance where the curves of all the rays become
straight lines, that the relative luminosities of the
different rays in spectra of ordinary intensity are practi-
cally the same. In the experiments last described, the
D light on the screen when not reduced by the sectors
was equivalent to '027 of a candle at 1 ft. This
would bring it far beyond the point where its curve,
and indeed those of all other rays, would become
straight lines.^

1 It must be remembered that we are only dealing with light reflected
from a white screen, and it does not follow that the lines may continue
straight indefinitely when the light is of the brilliancy seen when looking
direct at a bright spectrum, such as that of the sun, with a fairly wide
slit to the collimator.

THE MEASUREMENT OF LUMINOSITY 103

The following table shows the agreement of the
results of these last measurements with those of the
observations, from which the luminosity curve for the
central part of the eye was constructed. The quotient
of the difference of two abscissse in the straight part of
each curve divided by the difference of the correspond-
ing ordinates evidently is the tangent of the inclination
to the vertical, which, as stated above, is a measure of
the luminosity of the corresponding ray. In Column IV.
of the table, the first five of these quotients are multi-
plied by 28*2 in order to make the maximum luminosity
100. In the case of the last three entries in the table,
the beam of white light had necessarily to be diminished
in intensity before it passed through the sectors, and to
bring the luminosities to the same scale the tangents
had only to be multiplied by 5*03.

Table VII. — Relative Luminosities of Mays.

I.

II.
Wave-

III.

IV.

V.

Scale No.
56-3

Tanf^ent of

Tanji:ent

Luminosity of

Lengths.

Inclination.

X 28-2 or 503.

Normal Curve.

1

6358

1-5

42-3

435 !

50-6

5889

3-5

98-6

99-5

46-3

5618

316

- 88

88

39-3

5246

1-55

43-7

44-5

35-3

5066

•77

21-7

20-2

33-3

4975

2-39

12

12

29-3

4822

•96

4-9

5

19-3

4497

'33

l-«8

1-5

Luminosity of Spectrum of Light of Lotv Grade.

Should the luminosity of the spectrum of artificial
lights of low grade be required, we have to proceed

104 RESEARCHES IN COLOUR VISION

somewhat differently. Let us suppose that we wish
to ascertain the luminosity of the spectrum of the
brightest part of the flame of a paraffin lamp. In
this case we form a smaller prismatic spectrum, using
perhaps only one prism, and do not use a camera
with its lens, but only a lens of slightly shorter focus
to throw the spectrum on a white card. This must
take place in a darkened room, and the top of the
spectrum must fall on a finely-divided scale of
^ mm. An image of the paraffin light is thrown
on the slit of the collimator, which it should fill ; a
thin knitting-needle, mounted on a small leaden base,
is placed near the screen in the spectrum ; a thin
strip in the spectrum is cut out and appears black.
A comparison light — that emitted by a part of one
of the legs of an incandescent light — illuminates
this shadow. The alterations in the brightness of the
comparison lights may be effected by the graduated
annulus which has already been described. The two
shadows thrown by the spectrum and the comparison
light by the intervention of the needle are made to
touch one another. The luminosity of the part of the
spectrum measured is that which touches the shadow
illuminated by the comparison light. Care must be
taken that the brightness of the spectrum is not of
such a nature as to allow it to come into the category
of a feeble spectra of which the relative luminosities
of the rays differ (as we have seen at the beginning
of this chapter) fi'om those where it is fairly bright.
By using a wide slit in the collimator and a short
spectrum, this can always be effected. The measures
are not quite so exact as when a colour patch is em-
ployed, but the mean of repeated readings will give
results which are sufficiently close to the truth.

THE MEASUEEMENT OF LUMINOSITY 105

Sum of Separate Luminosities Equal to the

Combined Luminosity.

In the early measures of luminosity,^ it was proved
by repeated measures that the luminosity of a mixed
light is equal to the sum of the impression of each of
the components. To test the illuminating value of
colour mixtures, three slits were placed in the spectrum,
in the red, green, and violet. The luminosities of the
rays coming through each were measured —(1) separately ;
(2) in pairs ; (3) the whole combined. The measures
then made were on a different scale of units to those
at present employed, but they are none the less com-
parable. ^^^^^^ ^^„^

Observed.

R

(R+G)

G

(G + V)

V

(R+V) .

(K + G + V)

203
242

38-5

45

8-5

214
250

Calculnted.

204-25

241-76

37-50

4600

8-5

212-5
250-25

Combining these together we get —

R + G + V = 260
(R+G) + V= 250-6
(R4.V)+G = 252-5
(G + V) + R=248
(R+G + V)=250

= 260*25 by least squares.

Various other measures with slits at different parts
of the spectrum were made, and all went to prove the
correctness of the assumption made.

Within the limits of error of observation the lumi-

^ See Paper No. 2.

106 KESEARCHES IN COLOUR VISION

nosity of the combined spectrum measured as white
equals the luminosity of spectrum colours measured
separately, the slit in the spectrum being of accurately
measured width. In order to make this measurement,
it became necessary to reduce the luminosity of the
recombined spectrum colours, so that the white reflected
beam which was used in measuring the separate spectrum
colours might be utilised.

A carefully graduated fixed sector with 10°, 13°,
and 5° double apertures was rotated in the spectrum,
and with these reduced intensities of the white patch
formed by the recombined spectrum a match was made
with the reflected white light of the colour patch
apparatus. The luminosity of the ray of maximum in-
tensity, SSN. 50, was also measured. Knowing the area
of the curve obtained by the measurement of the rays in
the diflTerent positions of the accurately measured slit H
(p. 41), the result, as already indicated, showed that the
area of the luminosity curve was equal to the luminosity
of the white of the combined spectrum.

[The above results were obtained whether the annulus
or the sector was employed, and whether the reduction
in intensity of the recombined spectrum white was
effected by glasses of different densities of black
placed in the spectrum or by the sector with fixed
aperture.]

Flicker Luminosity.

When obtaining the luminosity of different parts
of the spectrum as it appeared to those colour blind,
who could only undergo a brief examination, it was
suggested by Dr. Watson, F.R.S., that perhaps the
measures obtained by the flicker method might give
similar results to the luminosity method and probably

THE MEASUKEMENT OF LUMINOSITY 107

be less difficult to a person wholly untrained in making
observations. The flicker method is dependent on the
fact that when a colour and (say) a white are alter-
nately brought on to a screen, following one another
with great rapidity, there is a sensation of flickering of
the light. When one or other of them is reduced in
brightness, a stage is reached in which the flickering
gives way to a quiescence, and no real flicker is ob-
served. According to some writers, this absence of
flicker enables the luminosity to be determined. Thus
if a green be observed on a small screen for a fraction
of a second, and immediately succeeding it a white
is shown for the same length of time, and again the
green is observed to be followed once more by the
white, and so on, the probability is that the alter-
nations of colour and white will give a distinct flicker.
If by some suitable means the luminosity of the white
be increased or diminished, at some stage in the altera-
tion of the brightness of the white the flickering
will cease and the small screen will show a mixture
of the green and white in a state of quiescence. The
brightness of the white, when this occurs, is held to
give the luminosity of the green. We shall see shortly
that the brightness by the flicker method is not
exactly the same as that obtained by the shadow
method, but, as used by us, the former is capable
of conversion into the latter without any appreciable

error.

Flicker Apparatus.

Dr. Watson's flicker apparatus is shown in the
accompanying illustration (Fig. 36). It is so constructed
and placed that the beams of light do not overlap but
follow one another without any dark or overlapping
interval between them.

108 EESEARCHES IN COLOUR VISION

The figure will give an idea of the complete
apparatus as designed by him. S is the white square
of magnesium carbonate on which the colour and the
white light are alternately thrown. AAA is an iron
band attached to a disc of sheet-iron, D, extending round
half the circumference. BB is a similar band on the
other side of the disc, also extending round the other
half of the circumference, as showu. It will be noticed
that where one band ends on one side of the disc

the other begins, so that beams of light which fall
on AA and BB respectively can be caused to fall
alternately on the white square S- The arrows a and
h show the direction of the two beams. The disc is
caused to rotate round its centre, Sp (a spindle), which
is connected with a pulley {not shown), and this is
connected with a small motor, M, to which a brake,
K, is attached. A speedometer, R, also registers the
speed of rotation. The white light is admitted to the
A side and the coloured beam to the B side. It is
placed between the recombining lens and the screen.

THE MEASUREMENT OF LUMINOSITY 109

It is obvious that in using this flicker apparatus
the reduction of the luminosity of the white light
could not be effectively made by the rotating sectors,
since it itself would cause a flicker. Recourse was
therefore made to an annulus, described at page 72,
placing its slit, through which the beam has to pass,
in the path of the reflected beam of white light, where
the rays from the lens cross in forming the white
image of the first surface of the first prism. At this
point there is an image of the slit of the collimator.
The slit in front of the annulus is opened fairly wide
so as to include the whole of the beam.

When using this apparatus the rotation of the
flicker wheel should be of such rapidity as to speedily
obtain a cessation of flicker. Experience has shown
that revolutions of 560 to 600 per minute are speeds
which for a fair intensity of spectrum suffice in
the red, yellow, and green. When the less bright
portions of the spectrum {i.e. the blue and the extreme
red) are under measurement, the speed may well be
reduced to 400 per minute. There is another point to
remember, viz. that there is a zone of brightness in
which no flicker is seen. In the bright part of the
spectrum this zone is very small and is rarely above
2° of the annulus used, but in the weak intensities it
may be as much as 10°, or even as much as 20° when
the brightness is very small. For this reason measures
should include alternate observations made first with the
white too bright and next with the white too feeble for
the flicker to be absent. The limits of the ** non-flicker "
zone are thus determined and the mean of the readings
may be taken as the place of minimum disturbance.

The following is a curve obtained from a spectrum
in which the D light was one candle at 1 metre
distant from the screen. The spectrum itself was formed

THE MEASUREMENT OF LUMINOSITY 111

Table IX. — Comparative Flicker and ^^ Shadow ^^ Luminosities of the
Lighi from an \%-ampere current Horizontal " Positive" Carbon
of an Arc Light.

^_— . —

SSN.

Flicker
Luminosity.

Flicker
Luminosity.

Shadow.
Max. = 100.

•5

Max. = 100.

■ • •

Max. = 92-5.

64

• • •

62

3-2

2-96

2

60

8-6

7-86

8-7

58

23

21-3

21 6

56

47-5

43-9

48-3 1

54

70

64-7

70

52

86

79-55

84-7

50

975

90-2

96-2

48

100

92-5

100

46

91-3

84-35

95

44

78-5

72-61

85-3

42

66*2

61-23

72

40

52-7

48-75

56-1

38

38-7

35-8

41

36

24-3

22-5

27-5

.34

12-8

11-84

15-8

32

7-7

7-12

8-9

30

5

4-62

6-17

28

3-6

3-33

4-6

26

2-8

2 59

3-5

24

• • ■

■ a •

2-7

22

• • ■

...

2-16

20

• ■ •

. ■ .

1-76

It should be here stated that the results obtained by
the flicker method ^ are with most eyes difficult to obtain
satisfactorily beyond SSN. 26 towards the violet, the
width of the " non-flicker " space being very wide, and it
may be wrong to assume that the true non-flicker took
place exactly at the centre of such a band.

' It must be remembered that it is only flicker against white that has
here been measured. The flicker of a coloured ray against the different
rays of the spectrum do not give the same flicker luminosities.

CHAPTER IX

COMPLEMENTARY AND CONTRAST COLOURS i

When one colour, " optically " mixed with another colour,
makes white (or grey when colour discs are used), these
colours are said to be complementary to one another.
Before, however, colours can be said to be complementary,
we have to know the quality of the white they match
when mixed. The white of daylight or of the arc electric
light, we know, requires a certain amount of some blue
ray to be added to the yellow-orange of the " D " light
to match the white of either of these sources, but the
blue ray which has to be taken will not be exactly the
same in the two cases. When, however, the match is
complete, the two pairs are complementary to one
another, and they will be complementary whether one
or both be diluted with the white light which has to
be matched. If, however, we take an extreme case of
finding the complementary to the orange when the so-
called white light is that of a paraffin lamp, candle, or of
an ordinary carbon filament glow-lamp, we are met
with a difficulty. The hue of any of these lights can be
very closely matched by orange rays in the spectrum.
Evidently, then, in such a case there can be no com-
plementary to this orange. We know, of course, that
these lights contain all the spectrum colours, and
amongst them, of course, those of higher refrangibility,
such as the blue rays, but the general mixture of them
does not enable the eye to distinguish the light irom
the spectrum colour with any degree of exactitude.

1 Paper No. 11.

112

COMPLEMENTARY COLOURS 113

One way of ascertaining exactly complementary colours
is to place three slits in one spectrum, and make a
match in hue with the ray from a second spectrum
whose complementary is sought.^ The hue may be
readily obtained by mixtures when the slits are
placed at the red lithium line, the magnesium *' b "
line, and the blue lithium line, or when the third slit
is placed well in the violet. If the D light is matched,
the mixture will be slightly paler than the single ray,
owing to a certain amount of white light which is in-
herent in the green ray.^ (The cause of this white light
being found in the green ray will be gathered after
reading Chapter XV., in which the method of finding the
colour sensations throughout the spectrum is described.)
When the hue is obtained, the white light for which the
complementary is to be sought should be thrown on the
cube surface in the colour patch apparatus, and after
noting the width of slits which give the match for the
ray, the slits are brought into position again by means
of the scale, and are again opened or closed until the
match to the white is made. The widths of the slits
are again measured, and from these two sets of measures
the width of slits required for the complementary colour
can be calculated. Suppose we take an example of the
D light. The width of the slits were found to be —

Red Slit. Green Slit.

100 25

When the white, for which the complementary colour to
be matched was that of electric light, the following were
the width of slits used : —

Red Slit. Greeu Slit. Violet Slit.

250 100 110

* A convenient plan is to use the modified apparatus given at p. 44.
' White is also in the blue ray if that be used, though it is not found if
the third slit be in violet.

H

114 RESEARCHES IN COLOUR VISION

Making the red slit readings equal in both cases, we
get for the D light —

Red 8lit. Green Slit.

100 + 25

and for the white light —

Red Slit. Green Slit. Violet Slit.

100 + 40 + 44

The complementary colour to the D light is there-
fore —

Green Slit. Violet Slit.

15 + 44

The slits can be set at these numbers, and the comple-
mentary colour is reproduced, and this can be matched
with a ray coming through a single slit in the second
spectrum. When found, its scale reading is noted,
which if necessary is converted into the wave-length.
This is a more roundabout way than the following.
When two spectra are produced in the same apparatus
(the " scaling " of both being accurately made), the ray
to which the complementary is required can be thrown
from one spectrum on to one half of the surface of
the cube, the white, which the ray and its comple-
mentary when mixed are required to match, is thrown
on the other half, using, of course, the rod to make the
shadows touch just in the middle of the square surface.
From the second spectrum a ray may be thrown on to
that half which is occupied by the colour. By trial, the
particular ray which matches the white with it, altering
the widths of the slits when necessary, is readily found.
The scale number of the second spectrum is noted and
converted where necessary into its wave-length.

When the complementary to a colour for an artificial
light is required, the same procedure may be adopted,

CONTKAST COLOURS 115

using, instead of the white reflected beam of the arc
light, the artificial light to illuminate one half of the
cube's surface.

When only one spectrum is available, two slits may
be placed at varying distances apart (the distance apart
being measured by reading the scale number when the
D light passes through each slit). The slide is then moved
in the spectrum till a position is found by trial where the
mixed colours match accurately the white which is being
used. The scale number is read oflF, and the positions of
the slits in the spectrum is thus known. [Of course,
when the slide in which the slits are fixed is graduated
to correspond with the transparent scale, as it is in
the writer s instrument, the distance apart of the centres
of the slits can be read off without resort to the reading
of the D light passing through the slits. When one
slit is kept in a fixed position and only the second one
moved, the reading of the scale number when the match
is made determines the position the slits occupy in the
spectrum, if the scale number is once determined when
the D light passes through the fixed slit.]

Simultaneous Contract Colours.

When two colours are viewed side by side, as, for
instance, when two strips of different colours fall on the
surface of the cube from the colour patch apparatus,
both colours ma)'^ appear to be altered in hue. The
change induced is caused by the simultaneous contrast of
the two colours. When one of the colours is white, the
change in colour is most marked, as it changes its hue
in a remarkable and (to most eyes) unexpected manner.
When measuring or observing colours which are in
juxtaposition, it is sometimes difficult to determine

116 RESEARCHES IN COLOUR VISION

whether the hue is real or whether it is produced by
contrasts. Artists are well aware of the value of these
contrasts. If a very vivid red is required in a picture,
he will manage to place a green near the red, and this
brightens the colour. Contrasts with colour and white
are much more recognisable when the two do not exactly
touch one another, but are separated by a mixture of
the two. Suppose the colour patch is in use and that
one of the spectrum colours is on the screen, with the
white superposed. If a thin rod be placed in the paths
of the two beams, there will be one shadow illuminated
by the colour and the other by the white, and inter-
mediate between the shadows will be a mixture of the
colour and white. The white will show the contrast,
taking hues of very varying nature, according to the
spectrum colour contrasted with it. In the following
table an endeavour has been made to give names to the
contrasts as seen by a normal eye when the white is
(1) that of the arc light, and (2) that of gas light : —

Contrast Colours,

UncoDtrasted

Spectrum

Colours.

Red

{

Orange
Yellow
Yellow-green

Green

Blue-green

Blue ^ '

Ultra marine
Violet

Coutraat

Spectrum

Colours.

Cherry red
Scarlet

Terra cotta
Raw sienna
Olive green
Emerald green
Grass green
Blue-green
Signal green
Cyanine blue
Violet blue
Blue violet

Ultra marine
Violet

Contrast White
in Electric
Arc Light.

Green-grey

Bluish green-
grey

Blue-grey

Light blue-grey

Umber

Pinkish lavender

Light pink

Dark piuk

Salmon

Yellow ochre

Brownish yellow

Dark greenish
yellow

Raw sienna

Burnt sienna

Contrast

Spectrum

Colours.

Cherry red
Scarlet

Light red
Olive green
Apple green
Emerald green
Emerald green
Blue-green
Peacock blue
Prussian blue
Violet blue
Blue violet

Ultra marine
Violet

Contrast White
in Gas Light.

Green-grey
Sap green

Green-grey
Pinkish grey
Dark mauve
Pink terra cotta
Pink terracotta
Pink terra cotta
Salmon

Reddish yellow
Brownish orange
Brownish yellow

Raw sienna
Yellow ochre

CONTRAST COLOURS

117

All the contrast colours given by the whites are pale
colours, and by no means saturated. It is often asserted
that the colours in the white evoked by contrast with
the spectrum colours are the complementary colours
mixed with white. We shall show that such does not
appear to be the case.

Before proceeding further, it may be useful to record
the changes in hue which are evoked by contrasting
different colours together. The following table will give
an idea of the changes that take place : —

Original Colours.

Red

Orange

»»

Green

>»

Blue

f»

Violet

Green

Orange

>9

Blue

•>

Violet

Orange

Blue

n

Violet

Violet

Blue

Change due to Contrast.

»»

»>

Red becomes yellower
unaltered but

brighter
becomes more orange
„ orange
Green becomes bluer

olive
yellower
Oi auge becomes redder

greener

•»

)9

»

Orange becomes green-grey
Green unaltered but

brighter
Blue becomes greener
Violet no marked change
Orange becomes yellower
Blue becomes more violet
Violet becomes bluer
Blue becomes deeper
Violet becomes bluer

No marked change takes place in either

To obtain this table, observations were made with
the double colour patch apparatus. Slits were placed
in four places in the first spectrum and in the same
positions in the second spectrum, viz. in the red, orange,
green, and violet. The contrasts in most cases were
very marked, as could be seen by causing the same
colours to fall on a white screen outside that on which
the observations of contrast were made.

Reverting to the contrast colours on the white, the
following arrangement was at first made. Two separate
colour patch apparatus were employed, the receiving
screens (the faces of white cubes) being placed about
1 ft. apart. Later the experiments were made by

118 RESEARCHES IN COLOtIR VISION

utilising the double spectrum apparatus (p. 44), which
formed the necessary two spectra aud gave also the
white light required. We will call the left-hand spec-
trum No. I., and the right-hand one No. II.

With No. II. instrument the colour contrast was
formed jietween white and a spectrum colour. The
colour emerged through a slit placed in the spectrum
and forming a patch on the cube, and the white was
that reflected from the first surface of the prism. A
thin rod f in. diameter placed Id the paths of the two
beams caused two shadows to be cast on the cube, one
illuminated by pure white light and the other by the
spectrum colour. These were separated from one
another by an interval illuminated by a mixture of the
spectral colour and white light, and on each side of the
shadows the same diluted colour was to be found. The
appearance of the side of the cube (called No. II.) was
as below.

A was a stripe of white light, B of colour, c c c of the
same colour diluted with white. The intensity of the
^ ^ D sodium light thrown on the surface

was '5 of a candle at 1 (t. distance ;
the intensity of the other colours
can be obtained from the luminosity
curve at p. 94.

The patch of colour from instru-
ment No. I. was thrown on the face
Via 38 ^^ ^ second cube (No. I.) 1 ft. away

from the first cube, and was used to
match the contrast colour produced on A, Fig. 38.
The beam of white light, which was nearly equally
divided between the first and second spectrum by
means of the bundle of glasses (see p. 39), also fell
on the face of this cube. The intensities of the colour

CONTRAST COLOURS 119

and white could be altered at will ; that of the colour
by opening or closing the slit through which the
colour came, and that of the white light by rotating
sectors. By this means the dilution of the colour could
be secured. (It may be mentioned that the effect of
using a strip of the face of this last cube equal in width
to the width of A was tried, but no advantage over
using the entire surface of the cube was found.)

The method of procedure was as follows. With
instrument No. II. the colour to be used and the white
beam were thrown on the face of the cube No. II.,
the luminosities of the two being made as nearly
equal as possible. With instrument No. I. a colour,
which it was judged was nearly the dominant colour
of the contrast colour of A, was thrown on the face
of the cube No. I. and white light added. When a
match was perfected by slight changes in the colour
and in the intensity of the added white, the scale
number of the colour was read, from which the. wave-
length could be determined, and the relative luminosi-
ties of the white and the colour were measured. The
luminosity of the D light with a slit of known aperture
had been determined. Hence in repeating an observa-
tion it was only necessary to read the apertures of the
slit and of the sector.

It was found that a slight change in the contrast
took place after repeatedly shifting the eyes from the
one cube to the other. For instance, the contrast caused
by green appeared to lose a little of its red hue, de-
generating into a brown-yellow. To get rid of this
difficulty an artifice was employed, which appeared to
be completely successful. An ordinary box stereoscope,
with the lenses removed, was mounted on a stand, and
in such a position that when the left eye only saw cube

120 RESEARCHES IN COLOUR VISION

No. 11. , the right eye saw only cube No. I. Thus, the
right eye never saw the contrast colour, whilst the left
never saw the match. In this way, by alternately
changing the direction of the eyes to the two cubes,
a match could be readily made. When the match was
considered satisfactory, both eyes were directed to a
moderately weak white light, and, after a short interval
of time, turned to the two cubes, when, if the contrast
colour on the one cube and the mixed colours on the
other appeared to match accurately, the necessary read-
ings were taken.

Subsequently it was found more convenient to move
the rod placed in the [)aths of the two beams of the

instrument No. II., so that only one shadow appeared.
In Fig. 39 the stripe of white light. A, is shown, and c c
are spaces on each side of A illuminated by the colour
mixed with the white. (It is obvious that the stripe of
colour could be equally well isolated. ) The7*e is no differ-
e7ice in the contrast colonics created in the white hy this
2)lan, showing that the presence of the saturated colour
is not necessary to give the full contrast. This is a very
significant fact, and may help to throw a light on the
cause of the contrast. The following table gives the
results of both sets of observations, as the results are
the same.

CONTKAST COLOURS

121

Table X. — Diluted Background.

Colour Cont

rftsted with '
Wave-

\¥hito.
Luminosity

Colour Produced bj
Dominant

f Contrast.

Proportion
of White

SSN

length

in terms of

RRN

Wave-length

kjK3i.^ •

of

one Candle

oox^ .

of Contrast

to Colour.

Colour.
672

at a foot off.
•15

29-4

Colour.
483

Wl

iite=L

1

i> 1 r 57-9

^^ { 56-3

054

636

•22

29^8

484

057

Orange | ^^.^

612

•44

29-9

485

■066

598

•46

30^7

487

•070

Yellow { %r.

585
569

•50
•49

28
26-8

481
471

100
120

f 45-7

558

•44

51

610

•165

Green { 42-7

541

•33

51 ^4

598

165

138

517

•13

51

592

170

r33-6

499

•07

50-4

587

175

Blue 29

481

•023

50

585

200

124-5

466

■012

49-8

583

•250

Violet . . .

All violet

...

49^5

581

•300

It will be seen from the table that different and
representative parts of the spectrum were used, being
the red, yellow, green, blue, and violet, and that in every
case the contrast colours provoked in the white could be
matched by a single colour of definite wave-length when
diluted by white light. If the contrast colour caused
by the green were its complementary diluted by white
light, it should be by a purple, which requires a mixture
of red and blue, whereas it is an orange. The fact as to
whether the contrast colour as matched could ever make
white when mixed with the colour which caused it was
very readily proved. The two colours were thrown on
the same cube, and the proportions of the colours altered.
In some few cases there was a very close approximation
to the formation of a white which matched the electric
light, but in the majority no match could be made.

Another set of experiments further exemplified this.
In instrument No. 11. three colours were chosen — one in

122 RESEARCHES IN COLOUR VISION

red, another in the green, and the third in the violet.
The same three colours were found in instrument No. I.,
and three adjustable slits placed in each of them. With
these three slits a match in the first instance was made
with the white of' the electric light— a contrast between
white and the red was then formed on the cube, illuminated
by No. II. instrument. The red was then shut off from
instrument No. I., and the mixed violet and green lights
were diluted with white light, but in no state of dilution
did the white stripe as coloured by contrast appear of
the same tint as the complementary colour of the red as
obtained from the diluted mixture. The same negative
results were obtained by making the contrast with the
greeu. With the violet a much nearer approach was
made.

This experiment was varied by matching the light
from an Argand gas burner, and forming the con-
trasts by means of the Bame quality of light. The same
negative results were again obtained.

The difference, if any, was next observed between a
contrast made by a saturated colour and that given by
the diluted colour.

In order to get a stripe of white enclosed between
two saturated stripes of colour, a Vernon-Harcourt screen
was employed instead of a rod (Fig. 40).
^ ° The principle of this may not be known

generally, so a brief description of it may
be necessary. It consists of a thin rect-
angular metallic plate of about two inches
wide, in which two broad slits, A and B,
Fio. 40. ^^^ ''"* ^^^ separated from each other by

C, the width of the slits. This plate, if
placed in the path of the beam, allows two stripes of
colour and of white to pass. By carefully adjusting the

CONTRAST COLOURS

123

position of this screen, a stripe of white may be enclosed
between two stripes of colour. The results are given in
Table XI.

Table XI. — Saturated Background,

Colour Contrasted with White.

Colour Produced by Contrast.

•

SSN.

Wave-lengfth of
Colour.

Luminosity in
Terms of

SSN.

Dominant Wave-
length of Con-

Proportion of
White to Colour.

57-9

Candle Power.

28^7

trast Colour.

White =1.

672

•15

481

•015

66-3

636

•22

301

485

•020

53-6

612

•44

30-7

486

•022

51 •S

598

•46

30^75

487

•024

50

585

•50

31-6

491

•025

47-5

569

•49

59-8

671

•035

457

558

•44

63^5

611

•052

42^7

541

•33

518

598

•066

38

517

•13

50-7

590

•066

33-6

499

•07

50

585

•066

29

481

•023

49^8

583

•068

24-5

466

•012

49^7

582

•070

All violet

493

1

580

•070

The contrasts with gas light, using the same light to
dilute a spectrum colour in instrument No. I., were also
measured, and these are given in Table XII.

Table XII. — Contrasts in Gas Light,

Wave-length of

Dominant Wave-

Colour.

length of Contrast.

636

485

585

590

568

598

499

592

465

589

All violet

588

There are such small diflTerences in the wave-lengths
of the contrasts produced by the diluted and saturated

124 RESEARCHES IN COLOUR VISION

colours that it may be presumed they are due to error of
observation, although each table is derived from the
mean of several observations extending over a period of
three years. It may be interesting to state that in
every case the extremes in the one series embraced the
mean value tabulated in the other series, and that in no
case did the mean differ from any single observation
more than X 2'5.

There is, however, a very simple means of noting the
accordance between the contrasts caused by the diluted
and saturated colours. With one instrument the con-
trast caused by the saturated colour can be shown on
one surface, and with the other the same colour, but
diluted, on another surface, so that the two can be
directly compared. To the eye the only difference
between the two was in the amount of dilution of the
colour produced by contrast ; otherwise they appeared
absolutely identical.

An endeavour was made to ascertain at the same

time what dark interval between the white and the

coloiurs would prevent the contrast being appreciable.

To do this a cube with a whitened surface was placed

^ as shown on the top of

another white surface with

a black interval between

the two (Fig. 41).

^ The colour patch was

'^ « thrown so as to fall only on

the cube c, whilst the white

beam illuminated the white

surface a as well. When the white beam was also

thrown on another cube a foot away it was practicable

to form an idea of the colour of a. The effect was

curious and interesting. When the black band h

CONTRAST COLOURS 125

was just J in. in depth, the eye being distant 4 ft.
from the cube, the white stripe A appeared strongly
coloured, a appeared very nearly white, and if by an
artifice saturated colour surrounded A, it was pure
white. If black intervals separated the white A from
the diluted colour c, the colour in A did not disappear ;
it appeared to be more diluted, but the colour still
remained. If, however, a black interval was on only
one side of A (that is, by placing the shadow against the
edge of the square and making the black interval
between the colour and A), when the colour was
saturated the white appeared perfectly white, whilst
if dilute just a shade of contrast colour was visible.

By placing a diluted coloured space in contact with
a pure white space which was in its turn in contact
with a saturated colour, it became possible with several
colours • to make the diluted colour appear white in
contrast to the contrast colour itself With red this
became impracticable.

CHAPTER X

NUMERICAL REGISTRATION OF COLOUR

It will be gathered that a colour is known when its
hue, its purity, and its luminosity are known. This
applies not only to the spectrum colours, but also to
the colours of objects in nature. It is to these last
that we will apply ourselves first. Suppose we have
to ascertain what is the spectrum colour which matches
a piece of brown paper. This can be done in the
following way. Place three slits in the spectrum, one
in the red, another in the green, and the third in the
blue. Fasten a piece of brown paper on half of the
receiving cube surface, and then illuminate it with
the white light in which it has to be viewed, and by
a rod or rods cut off the spectrum colours from it, and,
shielding the other half (a white surface) in the same
way, we can then make a match to the paper by opening
or closing the three slits. The apertures of these slits
are measured, and the strip of brown paper is replaced
by a white surface, which again is matched by the three
slits, and their apertures also measured.
Let the first be —

and the second-

Red. Green. Violet.
a + b + C

a' + y + c'

we can tell which colour is smallest in the first, and we
shall in this case find it is in the violet.

126

NUMERICAL REGISTRATION OF COLOUR 127

Taking the same proportions of each colour in the
first equation which exist in the white, we get the
brown paper colour made up of two equations : red,
green, and white.

Let a" h" be the proportion of red and green
necessary to make a white with c ; then the first equa-
tion becomes —

Rod. Green.

a'' + I/' + c = white,
and

Red. Green.

(a - a'') + (6 — f ) = brown paper, less the

white in a.

We therefore have the colour of the brown paper a
mixture of the red and green. Closing the slits of the
red and green to the apertures (a — a^^) and {h — V)
respectively, we make a mixture which matches the
brown paper in all respects except in its purity (mixture
with white). Using the double apparatus, we can move
a slit along the second spectrum, which will match in
colour the brown (less the white), and will be found
to be in some cases an orange (with other descriptions
of brown paper a yellow). The ray which matches the
colour of the brown paper ( — white) is called its domi-
nant colour. When extreme accuracy is required, it
may be that resort must be had to the methods in-
dicated in Chapter XVI. In that chapter it is shown
how from the equation alone the true dominant colour
may be arrived at.

Using a Single Slit to obtain the Dominant Colour.

Instead of using three slits in the spectrum, from
what has been said it will be seen that only one slit

128 RESEARCHES IN COLOUR VISION

need be employed together with the means of mixing
white light with the colour. In practice this is the
best arrangement for rapid determination of the domi-
nant colom*.^ The colour patch apparatus given on
p. 39 is employed, using the mirror G^" for obtaining
the white light which is required as an addition to the
colour. The slit is moved in the spectrum till a position
is found in which the hue is presumably correct. White
light from the plane reflected beam is then added until
the hue is proved correct nearly. If there is some
small inaccuracy, the slit is slightly altered in posi-
tion, when the match by means of white is again
tried. A few trials may be necessary to get a per-
fectly good match. When the colour is found, its
luminosity and that of the added white are measured,
and the colour of the pigment, or the light transmitted
through a coloured transparent medium, is then regis-
tered as a mixture of the dominant colour of a certain
luminosity, together with an added white luminosity.
The only case in which this plan will not answer is
when a purple has to be registered. There is no spec-
trum colour which can match such a colour with any
amount of white added. In such a case resort must
be had to the three-slit method, and the colour regis-
tered in terms of the complementary colour and white.
The writer had submitted to him for such registration
a certain number of signal glasses and coloured pigments,
and the annexed tables will give an idea of the use to
which this method may be put.

^ Paper No. 13.

NUMERICAL REGISTRATION OF COLOUR 129

Glass.

Railway Compan3r's red li^ht .
Another Company's red bght .

»

}f

ty

Railway Company's signal green

If

Maker's signal green
Bottle-green glass
Cobalt-blue gmss

i»

)i

Electric Light

Domi-
nant
Colour,
A.

Per-
centage

of
White

in
Colour.

Lumi-
nosity
(White
=100).

6250

7

10-4

6200

■ ■ •

10-4

6250

•••

9

4925

46

21-8

4925

38

16-2

5100

61

19-2

4925

24

7-6

5500

32

91

4675

38

4-4

Gas Light

Domi-
nant

Per-

centage

of

Lumi-
nosity
(Gas
light

Colour,
A.

White
in

Colour.

= 100).

6275

• . •

131

6200

12

13

6275

• • •

10

5070

60

18-1

5050

34

12-5

5170

62

19 4

5050

: 22

6-9

5320

50

10-6

4650

59

1
1

3 3

Colour.

Vermilion ....
Emerald green .
French blue
Brown paper

f, ff (greyer)
Orange ....
Chrome yellow .
Blue-green

Eosin dye (Sporting Times)
Cobalt . . . .

Dominant
CJolour, X.

6100
5220
4720
5940
5670
5915
5835
5005
6400
4820

Percentage of
White Light

2-5
59
61
50
67

4
26
42-5
72
55-5

Luminosity
(White=100).

14-8

22-7
4.4

25

19-5

62-5

77-7

14-8

44-7

14-5

It must be remembered that the above are colours
which vary in composition, and they are only given as
specimens of the manner in which they can be registered.

The pure colours of the spectrum cannot as a rule be
matched by the three-slit method, since, as we shall see
in Chapter XVL, there will be an excess of white in the
mixture compared with that of the ray which is to be
matched.

1

CHAFTER XI

COLOUR DISCS

The problem of the mixture of colours would be incom-
plete if no reference were made to the mixtures which
can be made by the rotation of colour discs in which
different colours are shown as sectors. It is proposed to
show that the same effect can be produced by passing
successive images of the colours rapidly before the eye as
if the colours were thrown one upon the other.

Let us place a cell containing a solution of a purple
colour such as permanganate of potash (to which no
single ray of the spec-
trum can make a match
even with the addition
of white light) in the
path of the white re-
flected beam. We may
place one slit in the red
of the spectrum and the
other in the blue, and by
opening or closing the
slits make an accurate
match of the purple
F,o 42. colour on the white sur-

face of the cube.
Let us next cut out a cardboard disc as shown in
Fig. 42, in which the angular apertures are all exactly
equal, and rapidly rotate it round its centre in front of

COLOUR DISCS 131

the two slits so that the red only passes through one pair
of apertures and the violet through the other. No rays
will pass through the outside and inside apertures at the
same time. On rotation it will be seen that the purple
has precisely the same hue as before, though of course
dimmer. The match will be apparent if a sector is
placed in the white beam. The effect of two colours
falling intermittently, but for equally short intervals,
on the eye is the same as when the intermittence is
absent and the coloured lights are mixed. This is due
to the persistence of the light on the retina, as explained
in Chapter III.

The image caused by the red light does not fade '
away before the blue image is impressed, and the com-
pound impression gives the same sensation of purple as
is given by the absolute mixture of the two lights on
the cube's surface.

For experiments in colour, this duration of impres-
sions is of great value. We can take advantage of it to
compound the colours of pigments together in a very
simple manner. For instance,
we can paint a circular disc
blue and red as shown, and by
causing it to rotate round its
centre a purple will be pro-
duced. A small electromotor
similar to that used for mak-
ing the movable sectors rotate,
having a bouche, screw, and nut
at the end of the spindle, will ^la 43

be found convenient for making

these experiments. The discs, perforated with a clean-
cut hole, can be slipped over the spindle, and rest against
the bouche. The nut will clamp the disc and cause it to

132 RESEARCHES IN COLOUR VISION

rotate with the spindle, and the colours on the disc will
then blend.

The motor shown in the figure will enable the discs
to be rotated sufficiently rapidly for the colours on
discs 8 in. in diameter to blend. It must here be statetj
that the brighter the light that is thrown on the discs.

the more rapid their rotation must be. Very convenient
discs for producing mixtures of colours by rotating discs
are a red (vermilion to which a trace of blue is added),
an emerald green, and a French ultramarine blue. Let
us call such a red R, the emerald green G, and the
ultramarine U. A white disc we will call W and a
black X.

A convenient diameter for the colour discs with such
an electromotor as shown is 6 inches, and for the black
and white 8 inches.'

> Small discs of, say, 2 inches diameter od thin card may be punted with
different coloured sectors, and if a pin be passed through their centres a
smart movement of the finger at the periphery will cause them to rotate
sufticiently rapidly to cause the colours to blend without flickering.

COLOUR DISCS 133

The discs should be of stout unglazed paper or of
thin card, and should present even surfaces of coloration.
Their centres should be pierced with a clean-cut hole the
size of the spindle of the motor, and a cut should be made
from the circumference to the centre as shown. This
enables the different discs to be inter-
locked. As many as five colours with
varying sector apertures can be shown
at one time. The white disc should
be a disc painted with zinc white, held
together by the minimum of white
gelatine or fish glue (see p. 46). The
amount of white light reflected from
the black (which should be ivory black and spread as
paint with the aid of the colourless size) must be de-
termined, the measurement being that given by light
falling nearly perpendicularly on the surface.

The examination of the colours of the discs must be
thorough if quantitative work is required from their
mixture.

In the first place, the measurement of the intensity
of spectrum colours reflected from the discs themselves (or
from portions of the painted paper or cards out of which
the discs were cut) must be made by one of the methods
given in Chapter VII. (pp. 79-84) ; and afterwards the
luminosity of the same colours can be measured directly
or can be calculated by means of the luminosity tables
(see p. 94). [Table XXXVIII. would have to be used
for finding the white impurity mixed with the dominant
colour. The dominant colour may not really be that
which matches the colour in the spectrum, for, as shown
elsewhere (p. 320), some colours have a yellower hue by
mixture with white.]

Before entering into the more elaborate measures to

134 EESEARCHES IN COLOUR VISION

which colour discs can be put, it is proposed to give their
simpler uses, and this will be done by an excerpt from
the writer's book, Colour Measurement and Mixture}

*'If we wish to produce a white, or rather a grey,
from three colours, we can take three small discs of
V (vermilion), E (emerald green), and U (French ultra-
marine) of equal diameter, and behind them place inter-
laced discs of black and white of larger diameter, rotating
the whole five on a common centre. We shall find that
by altering the proportions of the first three, we can get
a grey which can be exactly matched by a mixture of
black and white, X and W. It has already been shown
that even lampblack reflects a certain amount of white
light, so this amount of reflected white light has to be
added to the white in the outside sectors. In the sectors
used it was found that the following proportions of the
three colours were required : —

V = 124
E =143
U= 93

360

and to make the same grey it required

X=278
W= 82

360

Now the black reflected 3 '4 per cent, of white light, so
that really the proportions of black and white were —

X =268-6
W= 91-4

360-0

» S.P.C.K,

COLOUR DISCS 135

" These matches were made in the light emitted by
the crater of the positive pole of the electric light, and
are correct only for this light. The greys here are
dark greys, and such greys can be matched exactly by
throwing the white light in which the comparisons
were made on a white card [in a dark room] by means
of rotating sectors. We can prove whether our matches
are fairly correct from our previous measures of the
luminosity of these three colours in comparison with
that of white. The luminosities of V, E, and U, as
found from measures (given ante^ white being 100), are
36, 30, and 4*4. 124 of V would have a luminosity

-^^ or 12-4 ; 143 of E would have 11*92 ; and 93
ooO

of U would have 1*14, which added together give a

91*4
luminosity of 25*46. The luminosity of -w^r ^^ white

(which is from the mixture of black and white), comes
to 25*39, so that we may assume pur observations have
been fairly correct." If the rotating discs be moved
into any other light, the matching of the greys will
not be exact. Again, colours in the outer discs may
be matched with V, E, U, X, and W as inner discs.
The colour which has to be matched may possibly re-
quire X and W with it.

It may seem curious that X and W may have to
be added to the three colours in the inner discs, but
a little reflection will show why it is. Suppose we want
to know the composition of gamboge (Y) in terms of
the V, E, and U, we have a large disc of the Y and
also large discs of X and W. On rotation we shall find
that no U is required in the inner discs, and that the
general hue of the gamboge can be obtained by V and E
rotating. Mix these two in any proportions we like.

136 KESE ARCHES IN COLOUR VISION

we shall find that the mixture will never attain the
luminosity of Y ; consequently we must darken Y with
X. Even then we shall find that the rotating V and E
will always be a little less saturated in colour. This
means that on rotation V and E produce a certain
quantity of white light mixed with the yellow they
make. This necessitates adding some white to the
rotating disc containing X and Y, and finally we shall
get a match —

V E Y w X

172 + 188 = 75 + 45 + 240

This equation tells us one or both V and E are
impure colours containing white, and that they contain

45 + JJ23-4 ,3

between them at least = ttfk ^^ white.

360 360

Further, it tells us we can obtain the luminosity of

Y, as we know those of V and E as given in the previous

example, viz. 36 and 30 respectively, white being 100.

This makes the luminosity of the left-hand number of

the equation 17-2 + 15*67 or 32*87, and the right-hand

75
number - - Y+ 14*76. Consequently —

3G0

^-- Y = 32-87 -1476 = 18-11

— that is, the luminosity of Y is 86*9.

In an ordinary way we can find the luminosity of
a pigment of any colour by replacing it for either V, E,
orU.

Taking as an example an orange disc (O), the red
(V) had to be removed, and was replaced by O. A
match with the grey was made and found to be —

E u o w X

115 + 150 + 95 = 83 + 277

COLOUR DISCS 137

Knowing the luminosities of E and U, that of O is
determined —

115x30 150x4-4 95^_^ 3'4 x 277
360 ■*■ 360 ■*"360 360"^ 360

95 = 5319
or O = 56

These are examples of the simpler uses of colour
discs, but a further extension can be made by the
measurements which were suggested at the beginning
of this chapter. In the foregoing nothing is known
about the three colours employed, except that they are
scarlet, green, and blue, and all other pigment colours
are referred to them in that limited capacity. We
shall show in a subsequent chapter how discs can be
used to replace spectrum colours in a great many
instances.

When the illuminating source of light is a large
surface, such as the sky, the method before described
is more diflBcult to apply.^ It may be requisite for
some purposes to use such a source, as, for instance,
when one has to find a suitable coloured screen for
making a photograph giving the various colours of
objects as seen in daylight in their proper luminosities
in black and white. It then becomes necessary to
devise a plan by which rings of different colours can
be made of equal luminosity in ordinary daylight by
rotating them with the proper proportions of black.
The rings must be concentric and rotated round the
centre (see Fig. 46). The problem to solve is to ascer-
tain what amount of black ought to form part of each
ring to make the luminosities equal.

^ Paper No. 14.

138 RESEARCHES IN COLOUR VISION

In Chapter VIII,, p. 101, it is shown that oaly
one ray of the spectrum, a greenish yellow, progresses
in luminosity at the same rate as white light. Thus,
if part of a white screen be illuminated by this colour
and another part by white light, and the luminosities
are equal (say, to one candle), then if the two beams
are equally diminished they will still match in luaii-

Fio. *6.

3 i* the QQt of the spindle.

T Is a violet diso (methyl violet).

B ii a portion of a bine ring (French altraraarine).

R ., „ red ring (vermilion).

Q „ .. green ring (emerald green).

Y „ ,. yellow ring (chrome yellow).

W ,. „ white ring.

nosity until the light is so feeble that it ceases to
stimulate the retina. Other rays lying not far from
this ray, both on the red and green side of it, give
practically the same results. When, however, the red
is compared with the white, each being made equal
(say, to one candle), equal diminution of the beams will
not show the luminosities as remaining equal, for the red

COLOUR DISCS 139

becomes rapidly less luminous than the white. With
the blue-green, the blue, and the violet, the reverse is
the case, the white becoming darker than the colour
as the beams are equally diminished.

Further, it is shown in Chapter XII. that colour
disappears from all rays of the spectrum long before
{except in the case of the pure red) their light is ex-

Flo. 47.

Vr are jeUow digcB.
BX iB & black disc.

W „ white dUc

S is the nat of tlie spindle.

tinguished, this last owing to the feeble stimulation of
the retina. Naturally, as the colour begins to dis-
appear, the matching of the luminosity of the ray
under consideration with that of white will become
easier to carry out.

These facts make it possible to devise a ready
method of ascertaining the luminosity of any colour.

140 RESEARCHES IN COLOUR VISION

If we take two yellow discs, one (say) 8 in. in dia-
meter and the other 4 in., and between them
sandwich a pair of interlaced black and white discs of
G in. diameter, and rotate the four discs on a
rotating machine at a speed which will make the
black and white into a grey without scintillation, this
grey can be made, by altering the proportion of black
to white, to match the luminosity of the yellow. A
very exact match can be obtained by observing the
discs through a black transparent medium, such as
the black obtained on a photographic plate after
development with methol or amidol developers.
The deposit may be so dense that the yellow colour
may practically disappear, and the two dull greys
may then be readily matched. The luminosity
of the yellow in terms of the white is given by
the angle which the white subtends when the small
proportion of white reflected from the black is added
to it.

The same procedure may be adopted for a green
colour and its luminosity be obtained. Four or five
observations for each colour should be made.

When the luminosities of these two colours have
been determined, 4-in. discs of them may be inter-
laced with a blue, and a grey formed, which can be
matched with a grey formed of black and white as
before. From the angles which the sectors of the
colours subtend and of the black and white employed,
the luminosity of the blue can be calculated. The
luminosity of the blue being ascertained, a red disc
may be interlaced with the green and the blue disc,
and that of the red calculated. As a check, a black
and yellow disc may be interlaced and compared with
the colour given with the red and green discs inter-

COLOUR DISCS 141

laced, one of the pairs of course being of greater
diameter than the other.

To ascertain what degree of accuracy could be
attained, the following experiment is given in detail.
The light used was the arc light, and the measure-
ments as described above made.

It was found that the black reflected 3 "33 per
cent, of white light, and that when the luminosity of
the yellow was matched the interlaced black and white
discs occupied 82° and 278° respectively of the com-
pound disc. This gave the yellow a luminosity of 78,
white being 100. In a similar way, the luminosity of
the green was found to be 43. These two discs were
interlaced with a dark blue disc and a grey formed
which matched a grey formed by black and white.
The following equation was obtained : —

Yellow. Green. Bine. White. Black. White.

118 + 71 + 171 = 122 + 238 = 130

Yellow.

The luminosity of 118 = 1^ of 78 = 25-6.

^ 360

Green.

The luminosity of 71=^ of 43 = 8-5.

White.

The luminosity of 130=^ of 100 = 36-1.

Blue.

The luminosity of 171 is therefore represented by-

36-1 -(25-6 + 8-5) = 2

142 RESEABCBES IN COLOUR VISION
The lamiiKiaty of the Uoe {Mgment is thaefore-

??? Of 2 = 4-2
171

The lumiDonties of the three pgiiients were then
compared with that of white by the method described
at p. 87, and foond to be —

Yellow 77-7

Green 43-2

Blue • . . . 4-1

The Iwninosity of the blue only differs firom that
found by the new plan by O'l.

The red disc was then interlaced with the blue
and the green, and a grey formed as before, and from
calculation it was found that it had a luminosity of
32*5. Direct measurement made the luminosity 32*7.

Having obtained the luminosity of the three stan-
dard colours, that of any other colour can be calculated
by substituting for one of them a disc of such colour,
and again making a grey and matching it with a grey
formed by the black and white. This method can be
carried out in any light, whether candle light, electric
light, or daylight ; but the luminosities of the colours
will vary with the kind of light employed.

When the luminosities of the colours are determined,
the angles which the segments of the rings in Fig. 46
should subtend can be calculated after taking into
account the luminosity of the black employed.

(Each ring when rotated being equally luminous, an
appropriate screen, placed in front of a photographic plate,
will show equal density for each part of the developed
image of the disc. All objects photographed through
such a screen on similar plates will be rendered in proper
gradations of light and shade regardless of colour.)

PART II

CHAPTER XII

EXTINCTION OF COLOUR AND LIGHT ^

It is a matter of everyday experience that the colours
of nature are seen at their best when daylight or sun-
light is brilliant, and that as the day wanes most
coloured objects lose their colour and have a tendency
to become grey. After the sun has set, the colour of
light-hued flowers, for instance, rapidly lose the delicate
hues. A red rose will become almost black, green leaves
will become grey ; a scarlet-coloured brick will become
dark, losing its ruddiness. In these changes of the
quality of the rays reflected from the different objects,
there are evidently two phenomena which are more or
less mixed up together : (1) the loss of colour; (2) the
loss of light itself. Investigations into these phenomena
must evidently form part of experiments in colour vision
which have to be carried out in the laboratory, more
particularly when quantitative measures have to be
undertaken.

We again naturally turn to the spectrum as the best
source of colour, and light with which the scientific
investigation should be made. If the conditions under
which the extinction of either light or colour from the
different rays of the spectrum can be ascertained, it will
be comparatively easy to apply the results to the rays
reflected from coloured objects. In this chapter it is
proposed to start with the extinction of colour, and to
commence by describing some simple experiments which

* Papers Nos. 4, 7, 24, and 26.

144

EXTINCTION OF COLOUR AND LIGHT 145

will enable us to realise under what conditions the sense
of colour is lost.

Using the colour patch apparatus, let us place three
slits in the spectrum — one in the red, another in the
green, and the third in the violet. By substituting a
compounded lens in place of the re-
combining lens, we can form three
coloured patches side by side. (To
make this compound lens, it is con-
venient to take an ordinary spec-
tacle lens of the same focus as the
recombining lens, and to divide it
into three sections. As in Fig. 48, ""■ '"■,

a diamond can be used to cut the sections A, B, C as
shown. The three portions are then re-arranged and
mounted in a wooden frame, as shown in Fig. 49. The
three sections of the lens are placed as shown, the
thinnest parts of A and C being next to B in order to
have the optical centres of the three some small distance
apart.) Any of the three coloured images can be increased
or diminished in luminosity by adjusting the widths of the
slits. In the experiment to be
performed the red patch sboidd
be made decidedly brighter than
the green patch, and the green
pjij ^^ than the violet. Between the

slit of the colliinator and the
lens forming the image of the crater is placed the
graduated annulus.

The light getting to the collimator slit passes first of
all through the thinnest part of the annulus. The red
patch will still appear brighter than the green, and the
green than the violet. In the darkened room the annulus
is turned so that the white light forming the three

146 RESEARCHES IN COLOUR VISION

patches is gradually diminished. As the light gets
feebler, the red patch loses its luminosity more rapidly
than the green, and the violet also becomes enfeebled.
By turning the annulus, the light becomes so dimmed
that the red patch disappears altogether, whilst the
green and violet patches become dull and colourless.
Here we have an example of the total extinction of light
and colour in the red, and the extinction of colour only
in the green and the blue.^

Another striking experiment is to remove the slide
carrying the slits, and to place a lens of about 10-in.
focus in front of the collecting lens. This will form a
brilliant spectrum on a white screen (placed where the
square patches of light are generally received). Using
the graduated annulus as before, in front of the col-
limator slit, the spectrum can be gradually weakened
by interposing its thicker parts. The red will appear to
close towards the green, as will also the violet. When
the thickness interposed is still further increased, the red
will disappear altogether, and the yellow and blue will
follow, so that finally we have a grey patch left, where
with the bright spectrum there was bright green.

Another illustrative lecture experiment is to place
one slit in the spectrum about the red lithium line and
another near the E line in the green. The patch on the
screen will be a yellow formed by the mixed light after
adjusting the width of the slits. The annulus can again
be employed to diminish the light entering the slit of the
collimator until a grey patch is produced. The *'red"
slit can be closed altogether without apparently altering
the colour or luminosity of the patch. On increasing

^ This experiment illustrates what is called the Purkinje effect. It
need scarcely be said that all these experiments have to be carried out in
a darkened room.

EXTINCTION OF COLOUK AND LIGHT 147

the light, the patch will be otily the green component
of the mixed light. These experiments will have shown
qualitatively that both light and colour can be ex-
tinguished.

We will now proceed to show how the point of extinc-
tion of the colour of all rays first of all can be determined.
This can be best done by comparing a patch of white
with a patch of colour and reducing the luminosity of
the latter tiU it matches the white, when it also is
reduced to about the same luminosity as that to which
the colour has been reduced. A box, as shown in Fig. 50,
about 3 ft. long, is required. The lid can be removed,
so that if necessary S can be viewed with both eyes.
At one end of the box, shown in plan, is an eye-piece E.
The other end has at its centre a patch S, 1^ in. square,
whitened with zinc oxide, the rest of the inside of the
box being blackened. The monochromatic beam a coming
from the spectrum, and the reference beam &, are re-
flected by plain glass mirrors M^ and Mg to apertiu'es
T* and T'' in opposite sides of the box, and from just
inside these apertures, by right-angled prisms P' and P",
so as to fall on and cover S. Rods R^ and R" are inserted
in the box in the paths of the beams so that they illu-
minate opposite halves of S. Diaphragms inside the
box cut off any stray rays of light, and rotating sectors
placed at A and B regulate the strength of the beams.
[It has been found perhaps more convenient, instead of
the sector A being in the path of the coloured beam, to
have an annulus in front of the spectrum slit, and only
to have the sector B to control the white beam.] The
room containing the apparatus is darkened. The sectors
A are closed or the annulus is turned until no colour is
discernible in the monochromatic beam, whilst the in-
tensity of the white beam regulated by the sector B

148 RESEARCHES IN COLOUR VISION

gives the standard of whiteness to which the coloured
heam is to be reduced. It is worthy of notice that
when the white beam is entirely cut off, or made

KS{

very feeble, colour often seems absent from the mono-
chromatic light, but is again perceived when the
beam is brightened. This is especially the case with
the red part of the spectrum. The strength of the

EXTINCTION OF COLOUR AND LIGHT 149

coloured beam was therefore always reduced to the point
that no colour was apparent whatever was the strength
of the white beam. The apertures of the sectors (or,
if the annulus is used, its scale) are noted for each
colour. The direct measurement of the luminosity of
such a feeble light would be very difficult ; it was there-
fore determined in the following manner. The box and
sectors were removed, and a white screen was placed
at the same distance from M that S was. The slide
carrying the slit in the spectrum was also removed
so that a patch of white light was received on the
screen ; the luminosity of this was measured by direct
comparison with an amyl-acetate lamp. The mirror M^
was next removed, and the beam then fell on the screen
of the original apparatus. Its luminosity was then com-
pared with the reference beam. The slit slide being put
back in the spectrum, the luminosity of the D light was
measured against the same comparison light. The pro-
portion that the luminosity of the D light bore to the
recombined white patch was thus determined. As the
value of the white light reflected from M to the end of
the box was known from the first observation, the
luminosity of the D light so reflected was calculated.
The luminosity of the D light having been found, that
of all the other rays was calculated from the luminosity
curve derived from observations made with the central
portion of the retina (see Table IV.), as it was with
this part that the observations now being described
were made.

The actual value of each ray when the colour dis-
appeared was calculated from the aperture of the sectors,
or the scale of the annulus.

The two tables and Figs. 51 and 52 show the
luminosity of each colour of the spectrum to two difierent

150 RESEARCHES IN COLOUR VISION

If.

Table XIII.

III. IV.

I.

V.

Scale No.

X.
6957

Luminasity of

Spectrum of

Normal Brightness.

2

Redaction required

for Colour to

Difiappear when

D=l Candle.

•075

Redaction required

when every Colour

has a LuminodtT

of 1 Candle.

62

•0016

60

6728

7

•023

•00161

58

6521

21

•008

•00168

56

6330

50

•0035

•00175

54

6162

80

•0017

•00136

52

5996

96

•0014

•00136

50

5850

100

•0016

•0016

49

5783

99

•0025

•0025

48

5720

97

•0074

•0072

47

5058

92

•0061

■0056

46

5596

87

•0034

•00295

44

5481

75

•0027

•00202

42

5373

62-5

•0023

•00144

40

6270

50

•0019

•00095

38

6172

36

•0017

•00061

36

6086

24

•0018

•00043

34

5002

14-2

•0025

•00035

32

4927

8-5

•0036

00031

30

4848

5-7

•0049

•00028

28

4776

4

•006

•00024

26

4707

2^8

•0075

•00021

24

4639

2

•0105

•00021

22

4578

14

•0165

•00023

20

4517

M

•024

•00026

18

4469

•86

•032

•000276

16

4404

•7

•043

•000301

14

4349

•56

•054

•000302

12

4296

•45

•07

•000315

10

4295

•34

•096

•000332

8

4198

•26

•13

•000328

6

4151

•18

•17

•000326

4

4106

•14

•24

•000336

persons when the hue becomes that of the comparison
white. The amount of reduction^ for each ray is re-
corded which was required supposing the light of D had
a luminosity of one candle at 1 ft. ofF the screen. This

* In both cases the standard annulus was used for the reduction.

EXTINCTION OF COLOUR AND LIGHT 151

is shown in column IV. In column V. is shown the
reduction that would be required supposing each, ray

had a luminosity of one candle at 1 Jl. distant from
the screen.

The following shows part of a series of readings, and
is given to illustrate the closeness of the different
observations made : —

McsDor

Scot* No.

BMdines of tbe Aunului, in

Degree..

Reading, in

62-7

35 40 36

37

60

U6 135 130 130

135

57-36

11)7 1»8

197

0402

230 SBO S40

240

51-95

250 235 255 248

247

50-87

235 238 S50 S30 235

248

238

60-33

230 227 250 2.'i5 262

245

48-26

163 172 180 190 190

190

183

47-63

221 228 220

S23

46-47

230 243 245 250 225

846 256

241

152 RESEARCHES IN COLOUR VISION

Table XIV. — W.'e Cttrws.

I.

II.

III.

IV.

V.

Scale No.

X.

1

Luminosity of i
hpectruin.

Reduction required

for Colour to

Disappear when

D=l Candle.

Reduction required

when every Colour

' has a Luminosity

of 1 Candle.

62

6957

2

•056

•00112

60

6728

7

•014

•00098

58

6521

21

•005

•00106

56

6330

50

•0028

•0014

54

6152

80

•0017

•00136

52

5996

96

<X)J3

•00125

50

5850

100

•0017

•0017

49

5783

100

•0027

•0027

48

5720

97

•006

•00582

47

5658

92

•0032

•00294

46

5596

87

•0018

•00166

44

5481

75

•0014

•00105

42

5373

62-5

•0013

•00081

40

5270

50

•0014

•0007

38

5172

36

•0017

•00061

36

5085

24

•0021

•0005

34

5002

14-2

•0025

•00035

32

4924

8-5

•0033

•00028

30

4848

5-7

•0043

•000255

28

4776

4

•0052

•000208

26

4707

2-8

•0058

•000162

24

4639

2

•007

•000136

22

4578

1-4

•008

•000112

20

4517

11

•Oil

•000121

18

4459

■86

•015

•000129

16

4404

•7

•023

•000161

14

4349

•56

•033

•000186

12

4296

•45

•043

•000203

10

4245

•35

•054

•000189

8

4198

•26

•065

•000169

6

4151

•19

•07

•000133

4

4106

•14

•08

1

■000112

In the case of the whole series of readings, the D
light of the spectrum through the thinnest part of
the annulus was 0-145 candle at 1 ft. off the screen.
The mean readings were taken, and then, as before

EXTINCTION OF COLOUR AND LIGHT 153

stated, transformed into the result that would have
been obtained if the D light had been one candle at
1 fb. from the screen. Several separate series were
taken, and the mean of the means adopted for each scale
number.

[The tables and diagrams show that the reductions
in lumiuosity of the rays at each end of the spectrum

to match the white depends on the extinction of colour
in one or more of the three sensations, and sometimes on
the extinction of light in one of them (see Chapter XV.).]
In the red, beyond Scale No. 58, the extinction of the
colour of a luminosity of one candle at 1 ft. distant
from the screen is affected when the luminosity is
reduced to about 0'0016 candle, and the blue sensation
is extinguished when its luminosity is reduced from one
candle to closely 0*00009 candle. From observations
made by a red blind person, it was found that the

154 RESEARCHES IN COLOUR VISION

extinction of a green colour only stimulating the green
sensation was closely 0*0005 candle. The whole of the
spectrum rays were matched with white by this observer,
and in the green, of course, he matched a large portion
with full white, or with very slight reduction in luminosity.
In the results just given, no mention was made as
to the aperture which the patches of light subtended.
Evidently an inquiry as to the loss of colour had to be
made when the spot of colour subtended different angles
on the retina. Perhaps a definite case will show such
necessity. In moonlight the cherry-red of a brick wall
will first be visible when standing 6 ft. away ; but put
a red wafer on a black or white background, and the
red will have vanished if looked at the same distance
away. In such a case as this, evidently the angular size
of a coloured object has something to say to the resiJts.
The colour extinction box, when the writer made these
experiments, was abandoned, and a different method
employed, a dark room being used .instead of the
box. Two pieces of gla^s, each ground on both sides,
were placed nearly in contact, strips of paper keeping
them from absolutely touching. In front was a thin
black board with two holes, a couple of inches in dia-
meter, cut out, and the ground glass was placed behind
these two apertures. Behind the board, and touching
the ground glass, pairs of blackened thin brass dia-
phragms could be placed side by side. A very feeble
white light from the crater of the arc light was caused
to illuminate one of the apertures, whilst the colour
under examination filled the other. The white and the
colour were made of approximately equal luminosities.
The colour and the white were darkened together by
means of annuluses, and when the tints appeared to
match perfectly, the diminution required was taken

EXTINCTION OF COLOUR AND LIGHT 155

Tablb XV.

t

,1

U

Dlunetcr

Angular
Apsrture.

In Powen
art.

lU^idE

•^.

Riding

Lac.

H^Ma,

Log.

Annul ni.

Annuliu.

ADnulu..

1-42

0-94

l-S?' 0"

- -09

260

1-76

350

-99

300

0-724

I°30' 0"

- -48

345

1-89

335

112

280

1-59

0-525

l" 5' 0"

- -93

220

2-11

310

1-33

260

1-76

0-J5

43'43"

-1-52

210

219

295

1-46

236

1-98

0-17

21' 17"

-2-56

170

254

255

1-8

205

2235

oiies

10' 46"

-3-56

126

2925

210

il9

170

2-54

0-036

9' 9"

-4-81

75

3-355

156

2-66

120

2-97

0-012

3' 3"

-6-4

10

3-91

100

3-U

60

3-46

156 RESEARCHES IN COLOUR VISION

as the point at which the colour vanished. A large
number of different rays were examined with the centre
of the retina for colour persistency in this way, but the
following will suffice to show that the colour extinction
follows a definite rule as the aperture is diminished.

Nos. I. and II. are the same ray (Scale No. 44), but
with different intensities to commence with. No. I. was
measured by the writer, and No. II. by another observer.
No. III. was read by the writer, and was D in the
spectrum or Scale No. 50*6. It may be remarked that
with the small apertures the extinction of colour in the
red was impracticable, as the extinction of light and
colour took place together, as it should do according to
other experiments.

The intensity of the light to just cause a loss of
colour may be increased tenfold when the aperture is
diminished to one-eighth the diameter. In the ex-
tinction of light, we shall see presently that the same
increase in intensity only requires a diminution to one-
quarter the diameter.

This seems to show that the stimulus required to
produce colour is of a different order from that required
to produce light.

The Extinction of Light for the different paHs

of the Spectrum.

The next problem relating to this part of the subject
is the measurement of the reduction in intensity of
radiation, in order not only to extinguish colour, but
also to extinguish any sensation of light. In these
observations, the greatest care must be taken to obtain
the most sensitive condition of the retina. It is useless
to attempt any serious readings until the eye has been in

EXTINCTION OF COLOUK AND LIGHT 157

darkness for at least twelve minutes, if it has been previ-
ously saturated with ordinary daylight. The following
table shows readings made by the writer in extinguishing
light after the eye had been immediately withdrawn
from the daylight of the laboratory. The increase in
sensitiveness seems to follow a hyperbolic curve where
the times of reading are the abscissae and the ordinates
the extinction reading.^

Table XVI.

Times of Observation.

At the commencement
After 38"

53''

1' 11"

1'44"

2' 43"

3' 44"

4' 62"

6' 60"

6' 41"

r 28"

8' 32"
10' 46"
12'

Readings.

1

3-2

4-9

6-9
10-6
17
27-6
43
63
78
89
96
103
103

The extinctions are the reciprocals of the readings.

It is obvious that the best theoretical plan of making
observations would be to cut off all the light, and then
gradually add small intensities little by little until the
sensation of light was felt. By this procedure the eye
remains in its most sensitive condition. Practically,
this plan does not commend itself for adoption entirely.

^ It mast be remarked that in all these researches the time occupied
in darkness was frequently more than two hours. The readings of the
extinction of light of the spectrum were repeated two or three times, and
the only light that reached the retina was that of the very feebly lighted
spot which had to be extinguished.

158 RESEARCHES IN COLOUR VISION

It is found that when extinctions have to be made, the
eye should see a very faint glimmer of light, which
gradually has to be reduced till the sensation of light
has gone. The eye becomes very difficult to control as
to the direction of its axis when there is nothing which
can fix it. When a search has to be made for the
advent of the first small glimmer, it very often occurs
that the reading is rendered useless firom the fact that
the eye has wandered during the observation. If the
light is kept feeble in the first instance, the sensitiveness
of the retina is not sensibly impaired, and concordant
readings can be readily obtained.

Extinction is occasionally rendered difficult from
"intrinsic" light in the eye, even when it has been
kept a long time in the darkness. Sometimes there will
appear to be a flash of light exciting the whole of the
retina, such as may be felt when pressing the eyeballs.
Whether these flashes are due to blood pressure or some
other cause, it is not for a physicist to say. They are
absent apparently when the health of the observer is
good and the mind at rest. This intrinsic light has to be
discounted, but when a number of series of observations
have been made, the observer will soon know whether
he is observing a flash, or whether he is making a true
observation of extinction.

The original apparatus the writer employed was
usually of the form described below, but variations in its
arrangement and in the methods of observations were
made from time to time, in order to track out any
possible source of error.

BB (Fig. 54) is a closed box 3 ft. long and about
1 ft. wide and 1 ft. high, having two circular aper-
tures \\ in. in diameter in the positions shown. The
aperture at the side is covered on the inside by a piece

EXTINCTION OF COLOUR AND LIGHT 159

of glass, a, finely ground on both sides, and a tube, T,
is inserted in which diaphragms, D, of any required
aperture can be inserted. E is a tube fixed into the
other aperture, and should for comfort be fitted with an
end shaped to receive the eye, as the observations are
made through it. S is a cardboard screen inserted from

I

¥l(3. tii. — Apparatus to Measure Extinction of Liglit.

the top of the box, the aperture being rendered light-
tight by a batten. The screen is black except one
circular patch, which can be altered at pleasure in colour
or size, but which in the experiments now to be described
was white and I in. in diameter.

When using this Instrument the beam to be ex-

160 RESEARCHES IN COLOUR VISION

tinguished was directed through the tube T and dia-
phragm D on to a doubly-ground glass by which it ivas
diffused. A portion of the diffused beam was reflected
by the mirror M to the white patch on the screen at S.
By altering the diaphragm D, the amount of light falling
on S can be varied at pleasure, and it can be still further
regulated by putting the rotating sectors in the path of
the incident beam outside T.

The point of extinction was observed as follows.
The slits of the collimator and of the slide were closed
to convenient widths, and the light was subsequently
diminished by inserting diaphragms. Two methods of
extinction were tried : ( 1 ) The slit traversing the
spectrum was moved until the ray was found which
was just extinguished with each diaphragm ; and (2)
after placing the slit in fixed positions in the spectrum
at a known ray the light was diminished by the rotating
sectors as well as by the diaphragms. The latter is
evidently the more convenient plan, but both were fully
tried in order to determine whether the method of
reducing the light by the rotating sectors could be
relied on in experiments of this nature. The agreement
between the results obtained, which was as close as
could be expected in such experiments, convinced us of
the trustworthiness of the latter method.

The diaphragms used at D admitted the following
proportions of the original light to the screen S.

No. 0, ^; No. 1, xb; No. 2, ^J^; No. 3, ^4^;
No. 4, :f}js ; No. 5, ^ ; No. 6, ^^^ ; No. 7, ^V?t-

The method of diminishing the illumination of the
screen by ground glass was found to be most effective.
A beam of monochromatic light from the brightest part
of the spectrum can be diminished to such an extent
as to come within the limits of extinction by the rotating

EXTINCTION OF COLOUR AND. LIGHT 161

sectors, with the apertures of such an angular dimension
as to be properly read (say, more than 6°).

The P light coming through the spectrum slit was
measured against an amyl lamp (or candle) by placing a
white opaque screen at the aperture a (the tube T l>eing
removed). The luminosity of the D light being thus
known, that of any other ray could be calculated from
the curve A in Fig. 32. Another method of observation
was as follows. A diaphragm with a small circular
aperture was placed in front of the last prism of the

I

Fig. 55. — Metliod of Double Reflection into Extinction fiox.

colour patch apparatus. The patch of light on the
screen was now a small circular disc, instead of being
square, as before. A similar box was prepared to that
of Fig. 54, but the ground glass was omitted. The ray
of light now falling on M formed a circular patch on
the screen S, but the beam of light so formed is too
powerful to be extinguished by any readable aperture
of the rotating sectors ; it was therefore further re-
duced by placing in its path, and at an angle of 45°
to it, two parallel mirrors A, B (see Fig. 55). Each

mirror can be either silvered or plain glass ; three com-

L

162 KESEARCHES IN COLOUR VISION

binations of different reducing powers are therefore
possible, viz. : (a) both mirrors silvered ; (b) one plain
and one silvered ; (c) both plain.

The proportion of the light reflected with each com-
bination can be readily determined. When the last was
used, the intensity of c was almost exactly y^ of that
of a. As the rotating sectors gave a further extreme
reduction of, say, x^5> <* could be used of a manageable
intensity.

When employing this method, the collecting lens in
front of the spectrum was so adjusted that the recom-
bined beam from the whole spectrum formed a circular
spot on S, the position of the spot of light on S was
therefore the. same for all parts of the spectrum.

The absolute luminosity of the beam from D of the
spectrum was measured by placing an open screen at the
same distance from the mirror M (Fig. 54) that S was,
two silvered mirrors being used at A and B, and using
the amyl-acetate lamp for comparison. The absolute
luminosities of beji.ms from other parts of the spectrum
were then calculated from this by means of the lumi-
nosity curves.

The results obtained by using the rotating sectors
with this apparatus were also tested by the method
before described, and were found to be perfectly trust-
worthy.

From the observations made a curve was plotted
showing what was the proportion of the beam from
each part of the spectrum which was just not visible.
The absolute luminosity of each part of the spectrum
having been determined in the way explained above,
a second curve was plotted, of which the ordinates
represent the absolute luminosity of each part of the
spectrum at the extinction point, or, in other words,

EXTINCTION OF COLOUK AND LIGHT 163

the proportion which would be just not visible, supposing
that each part had been originally of the uniform
luminosity of, say, one candle. This curve rose from
the blue-green towards the red, when, after reaching
a maximum, it tended to drop again. There appeared
to be a similar irregularity at the violet end. It was
suspected that these irregularities might be caused by
some admixture of white light due to want of perfect
transparency of the prisms, and further investigation
showed that this was possibly the case, and that when
this stray white light was eliminated the curve became
of the form shown by the dotted line. Fig. 56.

A combination of " cobalt blue " and a " blue-green "
glass was used for the violet end of the spectrum,
and "stained-red" glass — i.e. glass flashed on one side
with copper, and on the other with gold — for the
red end.

The luminosity of each beam after passing through
the medium was determined, also the proportion left
when it was reduced so as just to extinguish the light,
the product of the numbers representing these quantities
would evidently represent the absolute luminosity at
the point of extinction, or, in other words, the propor-
tion left on the supposition of a uniform luminosity for
all parts of the spectrum.

The next figure shows the results of the extinction
of light in the different parts of the spectrum by an
eye neither myopic nor astigmatic, and which had normal
colour vision.

To the left of E there are two branches to the curve ;
the one shows the extinction when the central part
of the retina is used, and the other when the whole eye
is allowed to wander over the end of the box. As there
is no absorption by the yellow spot in these last,

1G4 RESEARCHES IN COLOUR VISION

they show that a greater diminution of the light is
required than after the rays pass through the absorbing

Fio. G6.— Extinction Curves at Nonnal Zje.

Tile contiDOOui line cnrres show the proportion of the beam from each part of
tbe spectrum which ie just notviaible, the illuminatloD by the beam from
D whon unreduced being equal to that of one amyl-aoetate lamp at I fc

The dotted onrrea show the proportion, aapposiag that all beams bad equal
intcDaitj to that of D.

medium. The tables ^ve the measures of each for
reference.

EXTINCTION OF COLOUR AND LIGHT 165

Table XVII.-

-Extinction

liy Ceniral Portion of formal Eye.

I.

II.

HI.

IV.

V.

VI.

R

'

Reduction of

L.

r> • I

Scale
No.

Wave-
length.

original

luminosity

in milHonths

to cause

Luminosity

of

original

beam.

ExL
100

Peraistency cun'e

E

(Maximum =100).

r

extinction.

62

7957

15,000

2

300

■ ■ •

60

6728

3,750

7

262*5

• • ■

58

6520

1,050

21

220-5

•62

'

56

6330

380

50

190

1-71

54

6152

196

80

156

3-32

52

5996

97

96

93-12

6-7

50

5850

35

100

35

18-6

48

5720

17

97

16-49

38-2

46

5596

10-2

87

8-87

63-7

44

5481

7-4

75

5-55

87-8

\

42

5373

6-55

62'5

4-09

99-5

40

5270

6-55

50

327

98-5

38

5172

6-85

36

2-46

95

36

5065

7-6

24

r82

81-3

34

5002

8-8

14-2

1-25

74

32

4924

11-6

8-5

•988

56

30

4848

16-3

5-5

•896

40

28

4776

26

4

1-04

25

26

4707

38-5

2-8

1-078

16-9

24

4639

56

1-82

1-019

11-6

22

4578

80

1-4

112

8-41

20

4517

107

1-08

1-156

6-1

18

4459

140

•86

1-204

4-64

16

4404

180

•7

1-26

3-6

14

4349

220

•56

1-232

2-95

12

4296

270

•45

1-215 1

2-4

10

4245

335

•34

1139

1-94

8

4197

430

•26

1-118

1-51

6 4151

510

•18

•918

1-27

4 4106

750

•14

1-05

1

-86

166 RESEARCHES IN COLOUR VISION

Table XYlll.—Exttnetion by Whole Eye.

I.

ir.

III.

IV.

V.

vr.

E.

Scale
No.

Ware-
length.

Reduction of

original

luminodtv

in milHontns

L.

Luminority

of

original

ExL
160'

Persistency curve
660

£

to cause

Wm.

(Alaximnm— 100).

extinction

•

38

5172

6-9

41-5

2-86

94-2

36

5085

7-4

33-5

2*48

87-8

34

5002

8

26-5

2-12

81-2

32

4924

8-8

21

1-85

73-8

30

4848

10

16-5

1-65

65

28

4776

11-5

13

1-49

66-6

26

4707

14-5

10-6

1-62

44-8

24

4639

18-5

8*2

1-62

341

22

4578

23

6 3

1-46

28*3

20

4517

30

6

1-6

217

18

4459

39

4

1-66

16-7

16

4404

51

31

1-59

12-3

14

4349

66

2-3

1-62

9-85

12

4296

80

1-9

1-62

8-12

10

4245

110

1-4

1-54

5-91

8

4197

154

1

1-54

4-22

6

4151

204

•75

1-54

3-18

4

4106

307

•5

1-54

211

2

4063

513

•3

1-64

1-26

4020

770

•2

1-54

•84

Column VI. in this and the previous table requires
a little explanation. It is called the persistency curve,*
and is derived from the reciprocals of the original ex-
tinctions, making the maximum 100. A little considera-
tion V7ill show that this curve is the luminosity curve of
the spectrum at the point of extinction. A comparison
of this curve with the luminosity curve of a feeble

^ Qeneral Festing and the writer gave it this name, though it has been
given by others to a different curve.

EXTINCTION OF COLOUK AND LIGHT 167

spectrum will show that the two are identical when the
maxima are made equal. We shall have to revert to
this curve in a subsequent chapter.

Modified Extinction Box. — In later experiments
made to test the effect the size of the spot had on the
extinction, a new form of extinction box was designed,
which for some purposes is more convenient than that
already described, since a graduated annulus, instead of
a sector, can be used with it.

gcT^;^vxvv,vv\Trvc^ \\\vv y i T \. ' CT.\^ ^ t.\xxx^^^5.v i'v

E

~i

B

\Vk\\^\^^'v^^x\<.N<skx».V<svvvVV«.vxV'wxV'vVVvvi.'vvvvVVv';

'^-x^,.

"-X.

^->H

<c-^

'^^V

-■■^'<=-.

'<^^^S>'

• /
I I
I I

4^

Fig. 57.

At the end of the box, B, an aperture was cut,
which was closed by a piece of glass, G, finely ground
on each side, or by an opal glass. Provision was made
for the insertion of diaphragms, c?, in front of the glass.
The eye-tube, E, was at the opposite end of the box, as
shown. Outside the box was a mirror, M, enclosed in
a frame, as indicated. The slit, S, in the spectrum and
the collecting lens, L, together with the annulus, W, are
shown in the figure. H is the handle used to move the
annulus round its axis. This form of box only admits
light through the end ; there is no reflected light in the
inside. The one thing necessary is to secure a good

IG8 RESEARCHES IN COLOUR VISION

scattering of light by the ground or opal glass, so that
the direct light is inappreciable compared with that scat-
tered, a desideratum which is obtained by using one or
two glasses ground on each surface. The box in which
M is fixed is blackened, or lined with black velvet, and
M itself can be either silvered or plain glass. In the
old arrangement the light entered from the side and
by reflection, and, after passage through ground glass,
illuminated a white disc at the end of the box. When
the disc was of fair size, any reflections from the black
interior were extinguished long before the light from
the disc itself vanished, and hence no inconvenience
was felt from the presence of the light from the black
interior, which may be taken as about ^ of that
reflected from the disc. If, however, the area of the
white disc is very much diminished, the illumination,
as will be seen presently, may be as much as 100
times greater than on the larger disc and yet be
invisible, and then the reflection from the interior of
the box would be visible after the extinction of light
on the small disc was completed. For this reason the
new form of extinction box was designed.

The first object in view was to ascertain whether
there is any difference in the extinction values of large
and small areas of light — that is, whether the images
on the retina are more speedily lost when the angles
they subtend are small than when large, and, if so,
whether the extinction of each separate spectrum colour
remains in the same proportion. For example, whether
the reduction in light necessary to produce extinction
of a 2-in. disc of colour at G, Fig. 56, is the same for
a disc of ^ in. or \ in. (the angular measures
of these when the apparatus described above is
employed are 4° 11', 1^ 3', and 31'), and, if not,

EXTINCTION OF COLOUR AND LIGHT 169

whether the reductions for every colour are pro-
portional. A series of observations made with these
and other discs showed that the smaller the disc the
less reduction in intensity of the ray was required to
extinguish it, and that the same ratio existed between
the extinction of the different colours. Fig. 58
shows graphically the results, which are in logs, of the
extinction value instead of the natural numbers. [This
enables the diagram to give the various curves without
having to change the scale value of the ordinates, as
was the case in the last diagram.]

There are five curves shown in all. No. I. is the
2-in. disc, No. II. the 'S-in. disc, No. HI. the •25-in. disc.
No. V. is the curve at p. 165 reduced to log. ordinates.
No. IV. is a pin-hole disc subtending an angle of 1' 29".
All these curves within the limits of error of observation
are parallel to one another, and as the ordinates are
logarithms this indicates that the rays are extinguished
proportionally. For curves L, II., and III. the D
light thrown on the ground glass, G, was '17 of an
amyl acetate (AL.) standard light at 1 ft. distance
(which is very closely '17 of a candle). With curve V.,
to which reference has already been made, the D light
was 1 AL. at 1 ft. distance. If the correction be made
for this difference of initial illumination, curve V.
would lie on curve I. With curve IV., where the
disc had a diameter of '012 in. and an angular aperture
of V 29'', the light had to be increased very largely
to enable any readings to be taken. It is inserted here
to show that with this small disc the curve is still
parallel to the others. To obtain the diagram, the
original readings were plotted' and free-hand curves

» For original readings, see Paper No. 7.

170 EESEAUCllES IN COLOUR VISION

EXTINCTION OF COLOUR AND LIGHT 171

drawn through them. The curves were made from the
following readings : —

Table XIX.

Standard

2-iDch

|-incL

^-inch

Scale

X.

carve.

disc.

disc.

diac.

Nos.

Curve V.

Curve I.

Curve II.

Curve III.

60

6728

36

• • •

• • •

• ■ •

58

6520

307

• • •

■ • •

• • •

66

6330

2-6

2-85

342

3-9

54

6152

22

2-4

2-96

3-4

52

5906

1^87

206

2 66

303

50

5850

1-64

1^76

22

2-72

48

5720

1-26

1-46

1-96

2*42

46

5596

102

1-26

1^76

2-25

44

5481

•87

1^1

1-67

2^09

42

5373

•82

1^06

1-52

2 02

40

5270

•82

106

1-62

203

38

5172

•84

107

1-67

207

36

5085

•88

1^1

re

21

84

5002

•96

1-2

1^7

2-2

32

4924

1^07

1-36

i-a3

2-36

30

4848

122

1^66

1-97

2-6

28

4776

1^41

1-62

212

2-62

26

4707

1-58

1-82

236

2-82

24

4634

176

202

2-62

2-97

22

4678

1-91

216

2-66

3 07

20

4617

2-06

2-28

2-77

3-2

18

4469

214

2-36

2-9

3-32

16

4404

2-26

2-6

3

347

14

4349

2-34

2 '67

316

3-68

12

4296

2-43

• • ■

• • •

• • •

10

4246

2-52

• • •

• • •

• • •

8

4197

2-63

• ■ •

• • •

• • ■

6

4161

2-76

• • •

• • •

• • •

4

4106

2-87

• • •

■ • •

• • ■

It has already been mentioned that some eyes have
a larger amount of colouring matter in the yellow spot
than others. The following table shows the difference
between the extinction of such an eye and one possessing
an ordinary amount of pigmentation. The light used in
this case was the arc light with the horizontal positive
carbons. The D light was 1 candle at 1 foot in each
case. The luminosities of the ray, multiplied by the

172 RESEARCHES IN COLOUR VISION

absolute extinction value, were practically the same in
each case, as would be expected.

SSN.

Pigmentatioii.

SSN.

Hgfmentation.

Normal.

ExoumTe.

Normal.

Excessive.

I

Log.

w.

Log-

i<«

1

60

1 3-64

' 3-65

30

•98

1-09

58

' 3-03

3

28

1-07

1-25

56

2-53

2-55

26

1-27

1-43

54

2-06

2-17

24

135

1-67

52

177

1-8

22

1-5

1-74

50

1-47

1-5

20

1-63

1-91

48

1-22

1-27

18

1-78

205 1

46

1-01

102

16

1-92

2-27 i

44

•87

•87

14

2-06

2-5

42

•83

•82

12

2-26

2-82

40

•82

•82

10

2^42

3

*

38

•87

•85

8

259

3-25 1

36

•88

•86

6

2-81

3-42

34

•89

•88

4

3

3-65

32

•92

i

•95

I

1

Law Connecting the Angular Aperture with the

Extinction.

The next investigation carried out was to ascertain
if any law connected the angular aperture of the object
observed with the diminution of the intensity of the light
which was required to cause invisibility. For the pur-
pose, a large number of diaphragms of very differing
apertures were inserted in front of the ground glass
(Fig. 57). For the sake of plotting, in the first instance,
and as they give the most rational scale, the diameters
of the discs were expressed in powers of 2, thus ^ inch,
which is 2"\ is used on the scale of abscissae as — 1 ; J^ as
— 2, and so on — all diameters not being expressed in exact
powers of 2 being calculated out in the ordinary way.

EXTINCTION OF COLOUR AND LIGHT 173

Tablb XX. — The following are the values j in inches^ of the apeiiurea
used. The table also gives the angles subtended and the valves in
poioers of 2.

Diameter in Inches.

Angles Subtenried.

Value in Powers of 2.

2

40 11' 0"

+ 1

1-5

3° 8' 0"

+ -6

-94

1° 57' 0"

- -09

•725

1^ 30^ 0"

- ^48

•525

r 5' 0"

- -93

•42

0° 52' 35"

-1-25

•35

0° 43' 43"

-1-52

•3

0° 37' 33"

-1-74

•17

0° 21' 17"

- 2-56

•086

0° 10' 46"

- 3-56

•036

0'' 4' 30"

-4-81

•012

0° 1' 30"

-6-4

These diaphragms were placed in front of the
ground glass, and the light from the discs thus formed
extinguished. In the first set of experiments, the pure
colours of the spectrum were employed ; whilst in the
others, ordinary lamp light and lamp light screened
with different colour glasses or solutions were used,
and identical results were found in all cases. The fol-
lowing figure (Fig. 59) was made from the mean readings
of some of the different series.

(The dotted line in the bottom curve was obtained
by calculation from the top curve.)

The indications here given are that the curves with
apertures less than 1^ in. diameter become straight
lines, all of which are parallel, and it is somewhat re-
markable that from that point the intensity of a light
which will be just extinguished with a certain diameter
of aperture may be increased ten times, and yet be
invisible when an aperture with one-quarter of that
diameter is employed ; if the intensity of the light be

174 RESEARCHES IN COLOUR VISION

increased 100 times, we have only to diminish the dia-
meter of the aperture to one-sixteenth, and it will again
disappear, or if to one-sixty -fourth, the light may be
increased 1000 times. There must, of course, be some

lower limit to this when the image, that of the small
disc of light, which a point such as a star subtends in
the retina. When the angular aperture exceeds 4°,
apparently the upper limit is reached, all extinctions
being the same beyond it.

EXTINCTION OF COLOUR AND LIGHT 175

Extinction dependent on the Least Diameter of

the Aperture.
In making extinctions of light, the observer must be
struck with the fact that before it finally disappears the

shape of the disc or object entirely disappears, and that
an irregularly-shaped spot is formed before it vanishes.
This phenomenon naturally leads to the query as to
whether within limits the shape of the illuminated spot
has any effect on the extinction. One of the earliest

176 RESEARCHES IN COLOUR VISION

experiments gave at all eveuts a partial answer to the
query. A slit was placed against the ground glass in
the apparatus (Fig. 57), and the extinction made with
lines of light varying between '0015 to '06555 in. in
width, so that the widest aperture was about forty
times broader than the narrowest, the length of the
slit being the same in all. The slit was illuminated
with white light or coloured light in the experiments
made. The extinction values were plotted diagram-
matically, the abscissae being the tvidths in powers of 2,
and the ordinates the logs of the intensities of radia-
tion, just invisible as before. Fig. 60 shows the results
with white light and red light.

The top slanting line is the extinction of a red light
and the lower one a white. The observations shoMr
that for every diminution in width to one-quarter, the
extinction value of the light may be increased ten times.

The table below shows the widths of the slits in

powers of 2, and the annulus readings converted into

logs.

Table XXI.

White Light.

1

Red Light

Absolute

Width in

Readings.

Absohite

Width in

Readings.

Width in

( Powers

-

1

Width in

Powers

- — _ _ _

Inches.
•06166

of 2.

Degs.
360

Logs.
•99

Inches.
•05.355

of 2.
-4-22

Degs.
192

Logs.

-4^02

2-35

•06365

-4*22

342

1-06

•04555

-4-46

176

2^45

•04565

-446

323

r23

•03755

-4-73

154

2-68

'03366

4-9

298

1-44

•02955

-5-08

135

2-84

•02966

-5^08

285

1-55

•02555

-5-29

124

2^94

•02566

-5-29

275

ves

•02155

-663

110

3-05

•02165

- 6-53

267

V72

•01755

-5-83

98

316

•01766

-5-83

260

1-85

•01355

-6^2

76

3-36

•01365

-6-2

228

2-04

' -00955

-6-67

50

3-57

•00965

-6-67

202

2-26

•00565

-7-49

166

2-67

•00166

-9-33

54

3-63

EXTINCTION OF COLOUR AND LIGHT 177

Taking into consideration the extinction curve of the
spectrum, and these results, we can see how the green
lines of a feeble spectrum will be the first to be seen
(perhaps colourless), whilst others, though present, will
fail to be seen except with a very wide slit.

A further experiment was made which confirmed the
previous measures. The extinction of the light from a
circular, a square, and a rectangular aperture of the same
area was made. The circular aperture had a diameter
•94 in., the sides of the square were '84 in., and of the
oblong 168 X '42 in. In addition, an oblong aperture
•84 X 42, exactly half the latter, was also used.

The following are the results of the extinction, and
in the last column are given the results that would have
been obtained from the curves already described : —

Table XXII.

Aperture.

! Circular disc, '94 in. diani.
■ Square, '84 in. side .
i Rectangle, 1-68 x -42 in. .
Rectangle, '84 x '42 in.

Width in
Powers of 2.

Read

ings.

Degrees.

Logs.

- -09

234

1 98

- -25

216

2-14

-1-25

152

2-69

-1-25

154

2-68

Logs from
Diagram.

1-98
215
2-65
2-65

Remarks on this table seem unnecessary, as they so
plainly indicate the guiding factor in the extinction.

This perhaps is one of the most curious results that
have been obtained, for it is hard to conceive that the
area of the retina impressed should not be a factor. The
experiments clearly show that the estimate of small
intensities of light by their effect on the light-perceiving
apparatus is not a simple matter. The extinction of
comparatively larger areas of light is most instructive.
The light from a square, or a disc, or an oblong, just

M

178 RESEARCHES IN COLOUR VISION

before extinction, is a fuzzy patch of grey, and appears
finally to depart almost as a point. This can scarcely
account for the smallest width of an illuminated surface
determining the intensity of the light just not visible ;
but it tells us that the light is still exercising some kind
of stimulus on the visual apparatus, even when all sensa-
tion of light is gone from the outer portions. The fact
that the disappearance of the image takes place in the
same manner, whether viewed centrally or excentrically,
tells us that this has nothing to do with the yellow spot,
or fovea, but is probably due to a radiation of sensation
(if it may be so called) in every direction on the retinal
surface. Supposing some part of the stimulus impressed
on one retinal element did radiate in all directions over
the surface of the retina, the effect would be greatest in
its immediate neighbourhood, and would be inappreci-
able at a small distance, but the influence exerted upon
an adjacent element might depend not only on its dis-
tance, but also upon whether it was or was not itself
excited independently. Following the matter out
further, we should eventually arrive at the centre of an
area, as the part which was the recipient of the greatest
amount of the radiated stimuli, and consequently that
would be the last to disappear. With a slit aperture,
the slit is visible till extinction is very nearly executed,
but it finally merges into a fuzzy spot at the moment
before it finally fails to make any impression of light.

Extinction of Light received Excentncally on

the Retina.

An investigation into the extinction of light at
different angular distances from the centre of the eye
was attempted. The experiments are of a very difficult

EXTINCTION OF COLOUR AND LIGHT 179

nature, and it requires long practice to enable a satis-
factory series to be made.

The method adopted was to place a pin, with a head
painted with Balmain paint, at every 5^ from the central
line joining the illuminated aperture and the position
occupied by the eye. The paint was very feebly phos-
phorescent, and only just sufficient to fix the centre of
the eye at the required angle from the object. The
results of two experiments, red and white light (paraffin),
at 10"^, are given. It appears from these that at this
angular distance the extinction of all light from the red
takes place when the light is about one-third brighter
than is required for the centre of the eye. With the
paraffin light it is somewhat less. With green light
about E, and with blue at the lithium line, the necessary
reduction of the light is greater than for the centre of
the eye, a result already shown. ^

Table XXIII.

Aperture.

Anglo.

2-«

Red Light.

White Light.

Direct.

10^ from axis.

Direct.

10* from axis.

•94

•724

•525

•42

•35

•3

•17

•086

o / //

1 57
1 30
1 5
52 35
43 43
37 33
21 17
10 46

•09
•48
•93
r25
1^52
1^74
2-56
3-56

275
252
225
217
195
185
152
93

255
230
204
195
174
162
125
75

305
270
265
252
235
215
174
118

290
265
240
230
220
200
157
105

There is a further falling-off of sensitiveness at greater
angles than those shown in the tables, but the extinction
is very difficult to make with certainty.

^ See Paper No. 4.

180 RESEARCHES IN COLOUR VISION

Kxtiiivtion of Light and Colour together.
At p. 155 the result of the extinction of colour iu
connection with the angular aperture was given, and
Fig. 53 shows the results diagrammatically ; but it
also shows the extinction of light for the same rays.
From the diagram we can see that there is an angular
aperture subtended on the retina at which any ray will
be extinguished both for colour and light at the same
time. The red ray is that for which the aperture will be
the largest.

Luminosity of the Light coming through different
Apertures.

As a side issue to extinction, the following observa-
tions were made, as having a bearing on the necessity of
keeping the two patches of colour and light equal in
making measurements of luminosity.

The point investigated, but without any great d^ree
of detail, was the comparative luminosities of the same
light coming through two apertures of different dia-
meters. The method adopted was as follows. The
ground glass was illuminated
uniformly with the light to be
tested, and two apertures cut in
a black mask were placed in con-
tact with it, as shown (Fig. 61).
Sectors were placed close behind
the larger aperture, and rotated
with angular apertures of any desired amount. In front
of the collimator of the colour patch apparatus the
annulus was placed so that a regular diminution of the
light could be effected. The sectors having been set at

^^5^^^^^^SET

EXTINCTION OF COLOUR AND LIGHT 181

90°, the light coming through the bigger aperture was
diminished to half. As the light coming through the small
aperture is extinguished long be/ore that coming through
the larger one^ there must be some intensity of light in
the extinction box when the two apertures will appear
equally bright to the eye. The light coming through
the slit is therefore diminished till the two appear
equally bright. The diminution of light is noted, that
coming through the larger aperture being diminished
twice as much as that coming through the smaller. The
sectors are again set at 45°, and the same procedure
adopted as before. In making these determinations, the
eye has to judge the brightness of very dissimilar sizes of
area, and it might be thought that this fact would present
an almost insuperable difficulty in making very accurate
measures. As a matter of fact, it was not so; the
greatest difficulty was encountered in those cases when
the light of the large aperture was so diminished that it
became colourless, whilst the other had very nearly its
original tint. The red was perhaps the hardest to judge
on that account; the other colours did not present
any great difficulty. One of the curious phenomena
encountered in these measures at times was a distinct
scintillation of the light emitted by the small aperture.
Sometimes this was perplexing, but never to the extent
to render the comparisons uncertain.

Fig. 62 shows the results when the two circular aper-
tures are '94 in. and '086 in. in diameter. The top
slanting line is where the illumination was by a blue
ray of the spectrum (SSN. 27*3), and the lower the D
sodium or SSN. 50-6.

What are usually abscissae are the log readings given
by the annulus, together with the sector readings (con-
verted into annulus readings), which rotates behind the

182 RESEARCHES IN COLOUR VISION

large aperture, and the ordinates are the logs given
by the annulus alone. The luminosity of the large
spot is equal to that of the small spot under these
conditions. If the luminosity of the spots were always
equal, no matter what size they were, the sector would
have to be at 180, i.e. not rotating, and the inclination

of the slanting lines would be 45°. As the sector is,
however, required, tbe inclination is less than 45°, as
shown in the figure, and this gives a value of the intensity
of the light at each spot when to the eye the luminoMty
is the same. The slanting lines are straight, and the
inclination Is alternatively determined by the extinction
values of the two apertures.

EXTINCTION OF COLOUR AND LIGHT 183

Table XXIV.

Sector in

Equivalent
Values

Saile K(

0. 27-3.

Scale No. 60^6.

Readingfs in
terms of

Readings

Readings in
terms of

Readings

Degrees.

Aunulus

converted

Annulus

converted

•

180

of Annuliis.

Scale.

into Logs.

Scale.

into

Logs.

S.

• • ■

L.

• • •

S.

• • •

L.

40

L.
40

S.
3-66

L.

• • •

3-66

90

36

90

125

3-23

2-92

110

145 '305

2-75

45

70

140

210

2-8

2^19

170

240 ; 2-54

1-94

22-5

105

200

305

2-28

138

240

345

1-94

1-03

11 •ss

140

230

370

202

•82

300 440

1-42

•22

5-6

175

295

470

1-46

•04

330 505

1-16

-•34

Extinction

200

325

525

1^21

-•51

340 540

t

1-08

- '64

S. and L. refer to the small and large apertures
respectively. From Fig. 62 it is found that the extinction
value of the large aperture, '94 diameter, requires 200°
more of the annulus to extinguish it than the smaller
aperture '086 diameter. This accounts for the last line
in the table.

Extinction of the Light in Spectrum Colmirs when
the Eye is not ** dark " adapted}

So far the experiments as to the extinction of any
sensation of light were made with a retina **dark*'
adapted, in which condition it is most sensitive. A
large number of experiments (not yet completely pub-
lished) have been made by the writer and others in
his laboratory on the extinction of the sensation of light
when the retina as a whole is subjected to illumination
by white or coloured light. For this a modification of

^ See Paper No. 27.

184 RESEARCHES IN COLOUR VISION

the extinction box was made, and so far as the experi-
ments themselves are concerned they may be accepted
as trustworthy. There is one factor, however, which
has not been taken into account, viz. the aperture of
the pupil of the eye. A difference in aperture will make
some difference in the amount of radiation of any ray
which is just too feeble to be recognised as light.

Box used in the Extinction of Light.

B is the box, as in Fig. 63. At the end of the box
is cut a hole ^ in. in diameter, and against it, inside, is

colour

Fio. 63.

placed a 4-in. disc of white matt paper, in the centre of
which is cut J-in. hole. Behind the box, a second end,
separated from the first by an interval of a couple of
inches ; opposite the aperture at the end of the box, is cut
another circular aperture 1 in. in diameter, against which
is placed a piece of doubly-ground white glass, and a
second piece can be placed behind this. The coloured

EXTINCTION OF COLOUR AND LIGHT 185

ray of the spectrum will form a patch on the ground
glasses, and the aperture in the end will allow the rays
to pass and be viewed through the eyepiece E, but it
will be limited to the ^-in. aperture cut out of the card.
The 4-in. disc is illuminated from the reflected beam or
other light through an aperture, CC, cut in the side of
the box. Diaphragms are placed in the box to limit the
view to the disc. At the side of CC is placed a small
disc which throws a black shadow J in. in diameter on
the large white disc. This is taken as a measure of the
blackness to be matched when extinguishing the colour.
An annulus or sector is placed in the white beam, so
that the luminosity may be reduced to any required
extent. Another " annulus " is placed in front of the
ray issuing from the slit in the spectrum. The white
beam which passes through the aperture at the near
end of the box falls on a dead black surface away from
the aperture filled by the ground glass. The whole box
is dead black. Such is the instrument which has been
used, and has been found effective for the purpose.

By closing CC the value of the extinction by a dark-
adapted eye can be carried out. The box is furnished
with a dark hood, so that the only light that reaches
the eye is from the end of the box. The shadow
thrown by the small disc at C will always be of the
same darkness, whatever intensity of light thrown
on the white disc may be, when all the 4-in. disc is
covered up except a white disc a little smaller than
the shadow cast.

It is not intended to give descriptions of anything
except the disappearance of the light coming through
d and its match with the black shadow, which of
course to the eye varies in blackness according to the
intensity of the illumination of the 4-in. disc.

186 RESEARCHES IN COLOUR VISION

Example of an Extinction with the Retina

illuminated.

In one case the luminosity of the white disc was
'2 candle after passing through an annulus at 20^ of
the scale. Each 25° gave exactly \ the illumination.
Measures were taken with the light passing through the
annulus at 20^ 70^ 120°, 170°, and 220°, the extinction
of light being made by another annulus in which every
degree corresponded to 0086 in logs.

The illuminations are for 20° .

j»

j»

99

>>

9)

)9

>»

9)

70°
120"
170°
220°

•2 candle.
•05 candle.
•0125 candle.
•00312 candle.
•00078 candle.

Table XXV. — I'ahle showing Comparative Extinction of the Sensation
of Light when the Retina is stimulated tpith diffei-ent degrees of
White Light.

asN.

X

KG'.

170*.

IW.

70\

80".

n*

GO

Log.

Inten-
sity.

Lo«.

Inten-
sity.

Log.

Inten-
sity.

Log.

Inten-
sity.

Log.

Inten-
sity.

6728

2-79

624

2-9

800

3

1000

3*05

1120

3-18

1

1,620

58

6520

2-25

178

2-36

230

2-49

310

271

513

2-88

760

56

6333

1*98

93

2-02

106

2-26

182

2-45

282

2-67

468

54

6152

1-68

48

1-85

71

2-11

129

2-36

230

2-54

347

52

5996

1*46

29

172

53

2

100

2-32

209

2-49

909

60

5860

i-2f;

18

1^63

43

1-94

87

2-28

190

2-49

909

48

5720

116

14 5

1-55

35-6

1-89

78

2-24

174

2-54

347

46

5596

1-07

35-5

1-46

288

1-85

71

2-28

190

2-67

468

44

5481

103

10-7

1*44

27-6

1-85

71

2-82

209

282

660

42

5378

•99

9-8

1-46

28-8

1-94

87

2-42

263

2-92

835

40

5270

•99

9-8

1-51

32-5

202

105

2-58

380

3^08

1.200

38

5172

1-03

10-7

1-59

39

2-11

129

271

513

3-18

1.520

36

5086

112

13-2

1-68

48

2-21

162

2-81

640

3-31

2.040

34

5002

1-23

17

1-8

63

2-36

230

2-92

835

344

2.750

32

4924

1-38

24

189

78

2-47

295

306

1160

8-59

3.900

SO

4848

146

29

202

105

2 58

380

318

1520

374

5.500

28

4776

1-65

35-6

211

129

2-69

460

3-3

2000

3-89

7,800

26

4707

1-63

43

2-21

162

279

620

3-35

2460

4

10.000

24

4639

1-72

53

2-31

304

2-88

760

3*48

3020

4-08

12,000

EXTINCTION OF COLOUR AND LIGHT 187

The intensities of the light in this table have to be
multiplied by 7-7 to compare it with the extinction
with the centre of the eye shown in Table XVII., where
the Bcale is in millionths, and the D light is equal to
1 candle.

It will be noticed that at SSN. 60 the red ray, when
the retinal illumination is ^2 candle, is extinguished

with an intensity 2*6 times greatei" than when the
illumination is "00078 candle, and that the ratio of
the maximum extinctions are for those illuminations as
9"8 to 309, or as 1 to 32. The observation recorded on
p. 148 as to the reappearance of the red when ex-
tinguLshing its colour is explained by these measures.
The white illuminates the retina more or less strongly,
and the red colour becomes visible.

An interesting and perhaps important fact is brought
out by these experiments. They show that as the white
which illuminates the retina is increased, the point of

188 RESEARCHES IN COLOUR VISION

maximum extinction travels between SSN. 40 and
SSN. 52. With a very strongly stimulated retina, the
point of maximiun extinction may lie nearer the red
than the latter SSN. To illustrate this the persistency
curves (reciprocals of the extinction curves) have been
calculated, making the maximum in each case 100.
The position of these maxima give the position of the
maximum extinction.

Tablb XXVI. — PersUtenry Curves of Extinction of Light on a
Retina differently stimulaied by White Light,

SSN.

22(r.

Dark adapted.

60 !

1-6

58

5-3

56

10

54

20

52

331

50

52-5

48

661

46

81-3

44

891

42

100

40

100

38

891

36

72-6

34

56-2

32

40

30

33

28

27

26

224

21

18-2

EXTINCTION OF COLOUR AND LIGHT 189

The extinction of the light depends on the size of
spot of light, and appears to follow the same law as
when the eye is dark adapted.

CHAPTER Xm

COLOUR FIELDS^

In Chapter 11. was given an illustration of the
colour blindness of the outside portions of the retina,
and perhaps it is as a type of colour blindness that
the phenomena is most interesting to the physicist,
though to the ophthalmologist a contracted field may
indicate something which helps the diagnosis of disease.
In this chapter the treatment of colour fields will be
entirely confined to results obtained with normal eyes
and with pure spectrum colours, the eye being dark
adapted. It will be seen in due course that the con-
sideration of the laws which, though empyric, govern
the extent of the fields, have something to say in
confirmation of the trichromatic theory of colour vision.

Colour Fields.

pPerhaps the first thing that should be explained
is what a colour field is. In the experiment cited in
Chapter II., the experimenter, in order to lose sight
of a spot of colour, was told to look at the spot and
then move his head to the right and left without
altering the direction of the eyes, and at a certain
angle which the axis of the eyes made with the
line joining the spot and the eye, the colour of the

^ Tn this chapter the luminosities are given in terms of an amyl-acetate

lamp (AL.)t which may be taken, as already stated, as the light of 1 standard

candle.

190

COLOUR FIELDS

191

spot disappeared. Had the experimenter been told
to move the head up and down, other angles would
have been found at which the spot disappeared.
Again, at other meridians, the same thing would
occur. If the angles were measured and a chart
made with the centre of the eye as the centre of
the chart, and circles indicating the angles from the
centre, and the meridians being indicated by lines
intersecting the centre, then these observations would
be charted and we should have a colour field.

Fig. 66.

In the figure we have such a chart, and we give
supposititious angles which the experimenter made in
moving his head. Suppose, right and left, the right
eye had to be turned 60^ and 40° respectively, and
up and down 50° each, and in the meridians at
45° on each side of the vertical axis, at 55° and 45°
and 50° and 42°, the chart of the field for the
ray, when it became colourless, would be as above,
being indicated by the dark thick line. The chart
might have been made slightly more complicated

192 RESEARCHES IN COLOUR VISION

by giving the angles as they would be projected on
the surface of a hemisphere. The different diameters
would also be increased in number, every 10° being
indicated. These are not the charts which we shall
use in this chapter, though the meridional angles of
the field will be increased so that every 30° are
shown.]

Apparatus for Testiiiy Colour Fields.

In order to obtain the colour fields of pure colour
special apparatus is required. Two forms were em-
ployed by the writer. The first was a perimeter of
ordinary form, but modified for use in a dark room.
The perimeter of the form employed is an instrument
consisting essentially of a semi-circular iron or brass
band which was graduated into degrees about 2 in. in
width, which can rotate round a pin or axis piercing
the centre of the metal band. There is a double chin
rest, on which, if the chin is placed in one hollow, one eye
is at the centre of the sphere, of which the semi-circular
band is a portion ; if the chin be placed in the other side
of the rest, the other eye occupies a similar position.
The diameter of the sphere is about 30 inches. To
adapt this for the spectrum colours, a mirror fastened
to a ball-and-socket joint is placed just below the
position occupied by the eye, i.e. just below the centre
of the sphere. By means of an arm the mirror can
reflect along the arc any beam of light falling on it.
The light reflected was so arranged that a circular spot
of any desired colour could be caused to travel along the
arc (which was covered with white) when it occupied any
angle with the vertical. The distance of the arc was
so arranged that the image of the first surface of the
first prism was in focus on it, and the spot was formed

■iqm-

COLOUR FIELDS 193

by placing a diaphragm against the prism. The in-
tensity of the colour could be altered — (1) by closing
or opening the slit through which the coloured ray
issued ; (2) by placing a graduated annulus in front of
the slit ; (3) by closing the slit of the collimator ; (4) by
using sectors in front of either slit. The mode of
observation was to cover up one eye, and the other eye
was at the centre of the sphere when the chin was on
the appropriate rest. A spot of coloured light was
caused to travel along the white band of the semi-circle
whilst the eye was directed to its centre, which was
marked by a pin point of Balmain's luminous paint.
When the colour of the light was judged to have gone,
the reading of the arc was taken. It was not very
difficult to cause one coloured spot and one white spot
to travel side by side, and this enabled an accurate
observation of the disappearance of colour from the spot
under consideration to be taken. This was usually
unnecessary, as the judgment as to the disappearance
of the colour without the comparison spot was very
accurate.

In the second form of perimeter, a hollow white
hemisphere made of "papier mAch^" was employed.
The centre of the surface was pierced with a circular
aperture some 1^^ in. in diameter. This aperture was
closed by a doubly-ground glass, and outside the shell
apertures of any desired shape or dimensions could be
placed in contact with the ground glass. The colour
patch apparatus was caused to throw the patch of colour
on to the ground glass. When the glass was removed,
the patch of white that the combining lens cast when
the whole spectrum was uncovered, fell upon the eye
when placed at the centre of the hemisphere. This
insured that every ray was fully received on the pupU

N

194 RESEARCHES IN COLOUR VISION

when the ground glass was again interposed. It may
be stated here, once for all, that when light falling
on the ground glass was measured, by placing a white
card in its place and balancing it with an amyl-acetate
lamp, it was found that the brightness of the ground
glass, as seen from the centre of the hemisphere, was,

90 45 40 IS iti m w k 4

\

40 /%

^)s^As9|o3|s40 45 50

1

1 1 1, ,■ .1.. . .1 , , , .

1

•

Fig. 67.

within a very small fraction, twelve times that which
was reflected from the white card.

The hemisphere was furnished with a chin and
cheek rest, which would move round a vertical axis.
It was divided internally into degrees. The eye was
directed to any part of the surface by means of a small
phosphorescent bead at the end of a stick ; and a small

COLOUR FIELDS 195

electric lamp, which could be switched on by a simple
movement of the hand, gave light sufficient to read the
position occupied by the bead at any desired instant.
The intensity of the light illuminating the ground glass
was altered by any of the four methods mentioned
above. The annulus was usually employed to effect
the alteration, and it could be rotated at the will of the
observer by a long handle attached to the rack and
pinion motion of the rotating gear.

Instead of the hemisphere a flat surface was also
used, as in Fig. 67. The circles were drawn as shown,
and the faint guiding light was moved along the dif-
ferent meridians, the colour being seen at the centre.
The chin-rest is shown. This method is very simple
and efiective.

Similarity of Fields for Different Colours.

It was essential to know whether the fields for each
coloiu* were of the same form when the illumination was
so adjusted that one point in a field of one colour
coincided with one point in the field of a different
colour. The following two sets of observations made
by the lyriter, and the succeeding ones by one of
his assistants (W. B.), will give the answer to the
inquiry.

An aperture of '525 in., subtending an angle 2° 30',
was inserted behind the ground glass, and the light
falling on the eye when D was the ray selected, was
4*5 amyl-acetate lamps, nearly equivalent to 4*5
candles, at 1 ft. (In future this light will be desig-
nated as AL., and this particular illumination would be
4-5 AL.)

The following rays were used to illuminate the

196 RESEARCHES IN COLOUR VISION

aperture : red lithium (X 6705), D (A 5892), a ray having
the standard scale number 36 (A 5085), and the blue

lithium ray {X 4603). These had respectively the
luminosities of '3, 4'5, 2'1, and "4 AL.

The measures were made with the right eye (see
Fig. 68).

-'-- ^-'

COLOUR FIELDS

197

Table XXVII.

Extent of Fields in Degrees.

Angle of Field
in Degrees.

KedU

D.

SSN. 36.
24

Blue Li.
40

35

36

30>

37

40

27

47

60

47

50

33

57

90

T.

55

57

38

65

120

51

53

36

60

150

41 ! 43

29

50

180

34

36

25

40

150^

35

36

26

40

120

37

38

27

45

90 .N.

40

42

28

49

60

38

40

27

45

30j

1

34

36

25

! i

42

T. is the temporal and N. the nasal side of the retina.

In the following observations the illumination by the
D light was much reduced, being only '23 AL., and for
certain reasons, which will be apparent, the ray at scale
number 417 was substituted for that at scale number 36.
The other three were the same as before (Fig. 69).

Table XXVIII.

Extent of Fields in Degrees.

Angle of Field

1 in Degrees.

Red Li.

D.
25

SSN. 41-7.

Blue Li.

23

15

28

30 ^

28

27

16

32

60 <

35

37

21

40

90 T.

38

40

23

47

120

35

37

22

42

150J

27

30

18

35

180

23

25

16

28

150]

25

26

16

29

120

28 .

30

18

32

90

N.

29

30

18

34

60

26

27

17

30

38^

23

25

16

28

198 RESEARCHES IN COLOUR VISION

Taking these sets of observations separately, the
diagrams show that the fields for properly selected
luminosities are evidently the same, the D and red

lithium being very close to one another. If we compare
the fields for the D and red lithium rays in the second
table with that of the field for the green (SSN. 36) in

COLOUR FIELDS 199

the first table, we shall see that they are practically
identical.

The next measurements were made by a different

person, and since, as before stated, his colour fields vary
considerably from the writer's, the confirmation obtained
by his measurements appears very conclusive. They

200 RESEARCHES IN COLOUR VISION

were made for illustrating a different part of the
research, but they will be given here and referred to
subsequently. Two places in the spectrum were selected,
such that the two rays when combined would give
white light, the white being that of the electric light,
which is practically indistinguishable from the sensation
produced by the coloured rays when falling on the
peripheral portions of the retina. The first positions
selected were in the red and green, at X 6500 and
X 5002, corresponding to the scale of the spectrum with
the numbers 57 '8 and 34. The relative luminosities of
the rays reaching the eye were 225 and 270 respectively.

Two other positions were chosen in the yellow-green
at X 5614, and in the blue at X 4603, corresponding to
the scale numbers of the spectrum 46*3 and 22*8. The
relative luminosities of the rays transmitted to the eye
were 96*5 and 2r5 respectively.

The colour field for each of these four colours was
taken with the left eye, and the following table shows
the residts (Fig. 70) : —

Table XXIX.

-

'

Extent of Fields in Degrees.

Angle of Field
in Degrees.

Red.

Green.

Yellow-green.

Blue.

30

36

39

36

30]

28

34

37

35

60

31

37

42

38

90

-N.

33

40

44

41

120

32

36

42

37

150 J

28 '

34

38

34

180

29 ,

35

39

36

160^

34

43

50

44

120

40

50

57

50

90

.T.

43

55

62

55

60

41

51

56

50

30j

33

1

38

43

39

COLOUR FIELDS

201

Here we have two fields, the green and the blue,
which are practically identical, showing that the limits
of the boundaries are not affected by the hue, though,
of course, the illumination is very different in the two
cases.

• Fields of Impure or Mixed Colours.

When considering the question of the fields of mixed
colours, such as those produced by pigments, it became
apparent that a crucial test as to their efficiency might
be made by mixing colours of the spectrum together to
imitate some single spectrum colour, and, after making
the mixture of the same luminosity, to compare the
fields. With this in view, a red and green, near E,
were mixed together to match the D light in hue and
in intensity. The fields for each colour, including D,
were taken, as also was that of the mixed colours.

The following table gives the results : —

Table XXX.

Extent of Field in Degrees.

; Angle of Field

inDegi

'ees.

Red.

Green.

P 4-R.

(Matching D.)

D.
38

0>,

35

36

33

30

35

35 33

36

60

. N.

.36

36 35

39

90

39

41 37

43

120

37

38 37

42

150

35

35 1 34

37

180

37

38 35

38

160 N

43

46 40

47

120

49

61 45

50

90 .T.

56

58 50

61

60

50

52 I 46

53

30j

39

40

35

42

202 RESEARCHES IN COLOUR VISION

These colour fields all have the same shape (Fig. 71).
They do not cut one another, and if we compare the

fields of the red and the green with those of the green
and the blue in the previous table, we shall see that
they practically coincide. Thus the fields of a red, two

COLOUK FIELDS 203

greens, and a blue are the same when proper luminosities
are taken for each. Before leaving this t-able, it is well
to point out that the field for D is considerably more
extended than that of the mixed colours, as are also
the fields for green and red separately. We may con-
clude that the intrinsic white light in each colour, when
added together, is greater than the intrinsic white light
in the D ray, which has been shown to be the case
in the chapter on colour equations. Colours of pig-
ments should therefore not give the same fields as the
spectrum colours with which they approximately match,
since they are impure colours.

Connection between Change of Intensity of Colour

and Extent of Field.

The difference in extent of field, caused by difference
in illumination, was next determined in the horizontal
directions. The four rays — red lithium, D, scale number
417 in the green, and the blue lithium — were experi-
mented with as being fairly representative of the whole
spectrum. The different rays were first allowed to pass
through the annulus at 0° ; and subsequently measures
were made after passing through it, when its readings
were 35, 70, . . . 280°, as every added 35° halved the
previous intensity. The D light coming through the
slit with the annulus at 0°, measured 4 '5 AL. The
following were the luminosities of the other rays coming
through the same slit : red lithium, '5 AL. ; SSN. (4r7),
3 '2 AL. ; and blue lithium, '3 AL.

204 RESEARCHES IN COLOUR VISION

4

1

Inteoirity'

Table XXXL

Degrees

Red Li

Reading of Horiiontal Field

in Degrees,
o. 41*7. Bine L

ithium. 1 D.

Scale N

ithium.

AnnuluB.

of Ray.

1

Tem-
poral.

54

.N««l. |3SS: Na«l.
42 57 45

Tem-
poral.

61

Nasal.
48

Tem-
poral.

43

Kaaal.
33

35

1 1

60

38 53 41

39

29

57

44

70

47

36 49 37

35

27

53

42

105

43

32 45 ! 34

32

24

50

38

140

-ft

39

29 41 31

28

22

46

34

175

35

26 37 28

25

19

42

31

210

^ '

32

24 33 26

21

16

39

29

245

12S

1

28

20 30 23

17

14

35

26

280

24

18 26 20

14

13

31

25

We find from the above that the average diminution
in field for each reduction of half intensity on the
temporal side is 3*75^, and on the nasal side close upon
3** (see Fig. 72). Using these figures, the above table,
would be as follows : —

Table XXXII.

Degfrees . 1 ntensi ty
Annulus. of Ray.

Reading of Horizontal Field in Degrees.

Re<i Utbium.

I

D.

, Scale No. 417.

Tem-
poral.

N""!- , ^{. I Na~l.

1

54

42

57

45

35

1

50-25

39

53-25

42

70

i

46*5

36

49-6

39

105

1

42-75

33

45-75

36

140

i

39

30

42

33

175

35-25

27

38-25

30

210

A

31-5

24

34-05

27

245

ISftff

27-75

21

30-67

24

280

24

18

27

21

1

Tem-
poral.

Nasal.

43

33

39-25

30

35-5

27

31-76

24

28

21

24-25

18

20-5

15

16-75

12

13

9

Blue Lithium.

Tem-
poral.

Nasal.

61

48

57-25

45

53-6

42

49-75

39

46

36

42-25

33

38-5

30

34-75

27

31

24

COLOUR FIELDS 205

With W. B. these numbers became 4 and 2-5
respectively, showing a consistent variation from our
own measures. With other eyes it may be expected
that the numbers will also vary, but it appears that

there is a diminution in the angle of field in an arith-
metical progression, as the intensity diminishes in
geometrical progression. The region of the macula
lutea was avoided in these observations, as it seemed
to be useless to attempt any observations on a part
of the retina which was evidently unsuited for them.

a06 RESEARCHES IN COLOUR VISION

Extent of Field fw the Different Rays
of the Spectrum.

Another set of experiments were carried out to
ascertain the extent of the colour fields for all colours

when a slit was passed unaltered through the specti
The following IS a table of three sets of obeerval

COLOUR FIELDS

207

taken by the writer. The two first were taken with an
aperture of '525 in., with an angular value of 2*^ 30'.
The third was taken with an aperture of "086 in.,
embracing an angle of 25' only, the temporal extent
being only observed with it. The luminosity of the D
light for each set of observations is given in the column
of remarks in the table.

Table XXXIII.

Scale
No.

No

. 1.

No

.2.

No. 3,

X.

Remarks.

Tem-
poral.

Nnsal.
34

Tem-
poral.

Nasal.

Tem-
poral.

62

6957

44

37

28

18

The luminosity of the D

60

6728

53

41

45

33

27

light in No. 1=45
All. ; an aperture of

58

6520

61

47

53

37

33

'525 in. was used at

56

6330

64

49

56

41

38

1 ft. distance.

54

6152

63

48

55

41

39

The luminosity of the D
light in No. 2=M

52

5996

60

46

52

38

36

50

5850

56

43

48

35

33

AL., with an aperture
of -525 in.

48

5720

52

40

44

32

29

The luminosity of the D

46

5596

49

38

40

30

25

light in No. 3 was

44

5481

46

35

37

28

22

•5 AL., an aperture of

42

5373

43

33

34

26

18t

•086 in. being used at

40

5270

40

31

32

24

16t

1 ft.
The readings marked t
were doubtful, as they

38

5172

38

29

30

23

14t

36

5085

37

28

29

22

13t

fell on or close to the

34

5002

39

29

30

23

13t

blind spot. They were

32

4929

42

32

33

25

16t

obtained by reading at

30

4848

47

36

39

30

21

a small angle to the
horizontal line.

28

4776

54

42

45

35

28

26

4707

61

47

52

39

.34

24

4639

65

50

56

42

37

22

4578

65

50

55

42

38

20

4517

61

47

53

39

34

18

4459

58

44

49

35

31

16

4404

54

41

46

33

29

14

4393

51

39

43

31

27

12

4296

49

38

41

29

25

10

4245

47

36

39

27

• • •

208 RESEARCHES IN COLOUR VISION

If we plot the curves from the above table, and take
the distaoce apart of the nasal from the temporal or-
dinateB, we shall Bnd that when the latter reads 40° the
former reads 30°, no matter what the colour may be ;
and that, when the field increases about 7^° on the tem-
poral side, the field on the nasal side incroEises nearly
6" — a variation which is in accordance with the table
showing the field with variation of intensities of the
beam (Fig. 72).

Dependence of Field on the Size of the
Coloured Spot.

In Chapter XII. it has been shown that the loss of
colour in the centre of the retina depends largely on the
size of the spot of light viewed. Such being the case, it

seemed probable that the boundaries of a field would
contract if the spot of light dependent on the aperture
used in the apparatus was diminished, and if so, it

COLOUR FIELDS

209

seemed possible that some expression might be found
which might connect the two together.

The same kind of perimeter was employed as before,
and the spot of light on the ground glai was duninished
in size by placing circular apertures of diminishing dia-
meter in contact with it. The fields were measured in
a horizontal direction only at first, and the following
table gives the mean of the actual measures. The in-
tensity of the D light was I'l AL.

Table XXXIV.

Diameter

of

Aperture

in Inches.

•94

•525

•36

•17

•086

•036

•012

Angle

Sub-

tended.

Diameter

of

Aperture

in Powers

of 2.

Red Lithinm.

D.

41«7.

BlueUtliium.

Tem-
poral.

Nasal.

Tem-
poral.

Nasal.

Tem-
poral.

Nasal.

Tem-
poral.

Nasal

4 18
2 30
1 34
49
25
10
3 30

- -09

- -93
-1^52
-2-56
-3-56
-4-8
-6-4

42
37
35
29
25
19

6.8.

32

28

26

22

17-5

14

12-5

48
43
39
34
29
25
20

35
32
29
25
21
18
15

38
33
31
25
20

6.8.
6.8.

28
25
23
18^5
15
12
9

50
47
42
37
32
27
20

37
34
31
27
23
21
15

6.8. is blind spot, where measures are impracticable.

This table, when plotted, gives a diagram (Fig. 74)
which shows that between apertures subtending 4*^ 28'
and 10' (the power of \ being taken for the scale of
abscissas), the fields decrease in extent and are practi-
cally straight lines. For each diminution in aperture
to \ diameter, the diminution in fields on the temporal
side, is 5°, and on the nasal side 4°. The diminution in
field for a diminution of \ the intensity of lights it will
be remembered, is 7*5° on the temporal side and 6^ on
the nasal side. The diminutions in field thus bear the
same ratio to one another, viz. 5 : 4. The diminution

by every J of the area is thus equivalent to \ of the

o

210 RESEARCHES IN COLOUR VISION

intensity of light. From p. 155, Chapter XII., this
might be expected, but the writer was by no means
prepared to find that the relationship could be measured
so closely. When the apertures used were greater than
the largest given in the table, scarcely any alteration of
the field was obtained. And it may be taken that to
the writer any aperture subtending more than 5° will
give the same field. And with apertures subtended
between 5° and 3°, the field will only slightly diminish.

CHAPTER XIV

THE THEORY OF COLOUR VISION

In the preceding pages it has been shown how the
luminosity of a colour can be measured, and the
luminosity of a bright spectrum has by this means
been ascertained. It has also been shown that the
luminosity of a spectrum, when of a feeble character,
fails to be able to stimulate the red sufficiently to com-
pare with the stimulation given by the other parts of
the spectrum, and that the maximum luminosity is no
longer found in the yellow, but is in the green, and that
the colours are all more or less degraded in hue, being
more grey than coloured, and finally, when the source
of light used for forming the spectrum or the spectrum
itself is dimmed, the last trace of light is to be found in
the green. Again, it has been shown that after all
colour has disappeared from the different rays a residual
light is left, and that by proper appliances this residual
light itself may be extinguished ; though, from the
nature of the experiments, some radiation still is extant,
though insufficient to stimulate the retina. Further, it
has been shown that the same absence of colour is found
when a fairly bright ray is received on a part of the
retina which is not central, and that for each different
ray we have a colour field the extent of which depends
on the brightness of the rays and on the size of section
of the beam which falls upon the retina.

These phenomena, together with pthers which are

2U

»

212 KESEARCHES IN COLOUR VISION

found in colour blindness, have to be explained by any
theory which is to be accepted. The physicist naturally
looks at the matter from a physical standpoint, and the
physiologist, equally naturally, regards it from a physio-
logical aspect. The true aspect must be that to which
both agree. The seat of colour sensation, whether in
the brain or on the retina, is an open question which
neither side of scientific thought has established. This
is a question which by-and-by will no doubt be settled,
but in the meantime the physicist at all events must be
content to utilise the hypothesis that the primary seat
and sensation is in the retina, which is an outcrop of the
brain. Mathematicians treated the mysterious ether, on
the oscillations of which our colour sense depends much
in the same way. The theory that is offered in theee
pages is the trichromatic theory of colour vision, which,
from the physicist's point of view, explains completely
the various phenomena met with. The trichromatic
theory was first propounded by Young, who was at
the time professor at the Royal Institution. He based
it on the postulate that there are three primary colours
in the spectrum, a primary colour being one which is
incapable of being matched by mixture of any other
colours, and that all the other spectrum colours could
be imitated by a mixture of two or three of the
primaries. In 1861 Clerk Maxwell took up the Young
theory, and was enabled by an ingenious apparatus
which he devised, to show by calculation from observed
measures the composition of the spectrum colours in
terms of the three arbitrarily chosen primary colours.
In the next chapter these observations and measures
will be discussed. It is right to observe here that
from his calculations he was the first to show that
the three colours need not necessarily be the primary

THE THEOEY OF COLOUK VISION 213

sensations of colour, but that stimulation of one or
more of the three sensations could account for all
colours.

Helmholtz followed Maxwell, and, as in all other
branches of science, he added largely to our knowledge
of the phenomena of colour vision. In his laboratory
Koenig worked out the form of the three sensations' curves,
indicating the strength of the sensations called into play
by the various spectrum colours. The writer next
attacked the problem, and published two separate sets
of curves.^ An account of the more recent determina-
tion of the sensations will be found in the next chapter.
The trichromatic theory, then, is a theory which recog-
nises only threQ colour sensations, and regards every
colour as the result of the stimulation of one, two, or
three of these sensations, and, it may be added, it can
also include what may be called the fundamental sensa-
tion of light. In the broad aspect of the theory, where
colour is of moderate brightness, this last is an \m-
necessary addition, as any effect the fundamental sensa-
tion may have is drowned by the greater brightness of
the colour. When the colours are not bright, as in a
feeble spectrum, the ftmdamental light has to find a
place in the theory. The theory reduces colours to
their very simplest form, and this is quite in accordance
with the method in which nature works. It is quite
open for other theories to be propounded in which
certain groups of colours in the spectrum are supposed
to be separately produced, but which fail when analysed
by mathematical considerations. Again, every spectrum
ray may be supposed to be a separate sensation, but
there is not warrant for such extravagance.

^ See Papers Nos. 5 and 6.

214 RESEARCHES IN COLOUR VISION

Hdmholtzs Sensation Curves.

Helmholtz suggested that every ray in the spectrum
affected each of the three sensations of red, green, and
hlue. His idea is shown in the figure. The top curve is
supposed to be the red sensation ; its height at various
parts of the spectrum is supposed to indicate the amount
of stimulus given to the sensation by each ray of the
spectrum. Similarly, Nos. 2 and 3 curves are supposed

Fig. 75. — Helmholtz's Colonr Sensation.

1. Red sensation.

2. Green sensation.
8. Blue sensation.

to represent the green and blue stimulation by the
different spectrum rays.

It will be seen shortly that Helmholtz's idea was
right in the main, though perhaps not quite exact in
certain details, when the subject is considered in the
light of modern researches.

The sensations which are excited must be due to
some action on sensitive apparatus which lie at the
base of the retina. It might be a mechanical action
or a chemical or an electrical action which causes the
sensation. It is most likely to be caused by a chemical
action, which, as we know, induces electrical action, and

THE THEORY OF COLOUR VISION 215

which is really a mechanical action from the molecular
point of view. How this action stimulates each of the
sensations is at present by no means settled. In any
case there must be some receiving apparatus in the
retina on which the light falls, and the energy of the
light converted into visual sensations. Be the appa-
ratus what it may, we have first to satisfy ourselves
that the impact of the spectrum on the retina can pro-
duce curves of sensation such as are shown in Helmholtz's
figure.

Action of Nonsynchronous Rays on the Sensation

Apparatits.

We can quite understand why a coloured ray can
cause a chemical decomposition of a substance in which
the rhythmic excursions of an atom or atoms from the
centre of attraction in a molecule are in exact tune
with the waves of light falling on such atoms. The
excursions may be so increased in extent by the
rhythmic energy supplied by the waves of light that
the atoms leave the molecule and give us a new mole-
cule. Possibly by the electric current set up, the
sensation of the colour is produced. But it is not so
easy to see why the rhythmic excursions of atoms in
the same molecule are also increased to the point of
molecular rupture when the wave-motion of the im-
pinging rays are not quite " in tune " with the rhythmic
excursions.

Photographic and Mechanical Examples.

K a sensitive salt, say the chloride of silver, be ex-
posed to the action of the spectrum, on development we
have a streak of reduced silver which varies in density

216 RESEARCHES IN COLOUR VISION

of deposit throughout its length. By careful measure-
ment of the opacities of the deposit at difierent poiuts,
and then referring them to a scale of graduation obtained
by developing a plate which has been exposed to known
intensities of light, we are able to make a curve which
shows the sensitiveness of the particular salt of silver
to the different rays of the spectrum. We have in

Fio. 76. — Effect of Speotram od Silver Chloride.

Fig. 76 the curve of sensitiveness of silver chloride, and
in Fig. 77 that of silver bromide to the different rays
of the spectrum. The place of maximum sensitiveness
is different in the two cases. If we mix the two salts
together, we should get a curve which is compounded
of the two curves. If a third silver salt had been
impressed, we should have a place of still different
maximum, and the curve of sensitiveness of the three
mixed silver salts would be one compounded of the

THE THEORY OF COLOUR VISION 217

curves of all three salts. If we can account for the
curve of sensitiveness of any one of the silver salts, the
reason of the curves of sensitiveness of any other salt
will be the same.

The fact is that the maximum of the curves show
the place in the spectrum where the vibrations causing
the ray are in tune with the vibrations of the chlorine

ITio, 77.— EHect of Spectram od Silver Bromide.

in the silver chloride, the chlorine being that part of
the molecule which is swung away and aunexed to
some other adjacent foreign molecule. We may also
take a mechanical example of the effect produced by
vibrations which are not in tune with, but have to act
on, a vibrating body. A simple apparatus, in which
two different pendulums are caused to act on one
another, one having a very light bob and the other
a heavier one, will be such an example. The first pen-

218 RESEARCHES IN COLOUR VISION

dulum may be taken as representing the chlorine atom,
and the other the ray of light. When the two pCDdu-
lums are of equal length and the heavy one is started
vibrating, the light one also begins to swing, and as it
is in tune with the vibrations of the first, the amplitude
constantly increases. Making the light-bob pendulum
a little longer or shorter than the other, and again
starting the swing of the latter, the lighter one com-
mences to swing. At first the heavy one will cause
it to swing with increasing amplitude, but by degrees
the two will begin to swing in opposite directions ; the
amplitude of the light pendulum will decrease and finally
come to rest, when it starts again as before. The
annexed figure shows the trace that the light pendulum

makes when acted upon by the heavy one. The increase
in amplitude is well marked, as is the period when it
comes to rest. Thua if the waves causing a ray of light
are out of tune with the atom's vibration, the amplitude
will still be increased ; and the increase can be such
as to swing the atom beyond the sphere of molecular
attraction, and so decomposes the molecule, but with
less facility than when the waves of light are " in tune."

liesonatoi' Curves.
Helmholtz has also shown that in the case of sound
acting on a resonator, not only does that sound which

THE THEORY OF COLOUR VISION 219

has the same period of vibration as the resonator set it
in vibration, but that sounds which differ slightly in
wave-motion from that which is sjmchronous with it
also- set the resonator vibrating. He shows that the
greater difference there is between the synchronous
sound and that applied, the less does the resonator
respond. The curve in which he shows the difference
in resonation is similar to that of those shown in
Figs. 66 and 67.

We see, then, that we may expect when the spec-
trum falls upon what we may call a visual receiving
apparatus, that not only will such apparatus respond
to the ray whose waves are " in tune " with it, but
that waves on each side of it will also cause it to
respond, though to a smaller extent, and that the
general shape of the curves would be the same as
found for a photographic simple salt. In some cases
we might expect that principal harmonics might also
give curves of lesser ordinates.

We shall find when the sensation curves are dis-
cussed in detail, as they will be in the next chapter,
that what has been supposed might be expected seems
to be found in one of them.

Dazzling Colours.

Before quitting the photographic simile, we may
notice what happens to a photographic image of the
spectrum when for moderate brightness a dazzling
brightness is substituted and the same exposure given.
Measuring the opacities of the different parts of the
developed image, we shall find that the top of the curve
is nearly flat for some distance on each side of the place
of maximum sensitiveness, instead of being a rounded
point. This flat top indicates that the silver salt has

220 RESEARCHES IN COLOUR VISION

been exhausted of its atoms, which are swung away,
and that the maximum decomposition has been obtained
by rays which are not '* in tune" with the atomic swing.
The effect of a dazzling coloured light should be
similar. All three sensations are stimulated by, say, a
green ray, the green stimulation being in preponder-
ance. If a dazzling green ray falls on a place in the
retina, we have the green sensation at its maximum
stimulation at once, and following quickly on we have
the red and blue sensations contained in the ray at
their maximum stimulation. When the three stimu-
lations are equal, the effect is to produce the sensation
of white. The green would thus appear nearly white,
with a slight tinge of green in it. From the sensation
composition of an orange ray, which is red and green,
we should find, on using the same argument, that the
dazzle colour of the orange would be a very bright
yellow of a hue in which the two stimulated sensations
are equal.

Visical Receiving Apparatus.

At the present time it is almost useless to discuss
the nature of the visual receiving apparatus, as opinions
differ in the physiological world even as to the functions
of the rods and cones in the retina. It may, however,
be said that to the physicist there is a strong inclina-
tion to believe that there is some substance or
substances attached to or inherent in retinal processes
which have the power of being altered by light waves.
The first thought is naturally that the visual purple*
might be such a substance, since it has been proved
that it bleaches in the light. Prima facie it has to

^ It is not found existing in some folly developed eyes which are
presumed to see colours.

THE THEOKY OF COLOUR VISION 221

be rejected on the grounds that its absorption spectrum
is that of a purple ; therefore it absorbs the green rays
and allows the red and blue to be transmitted. Where
there is such absorption as the visual purple possesses,
a chemical or heating action must take place, chiefly
in the green and but slightly in the red, yellow, and
blue, so that the effect of the green would be most
visualised, but the fact that every ray of the visible
spectrum is visualised, and that the yellow is most
luminous, makes it appear that we must look for a
more universally absorbing substance. A physicist
would have to look perhaps for some grey matter,
composed of triple molecules, which would absorb the
rays which evoke the three sensations. One molecule
might be of a nature to call forth the red, the second
the green, and the third the blue sensations, which
might be visualised by an electric current evolved as
the result of the chemical action. In case of complete
colour blindness, one of the three molecules might be
inert (as is the case in some cases of photographic
salts of silver, which become insensitive by special
preparation) ; or, in the case of incomplete colour
blindness, one might be less capable of chemical
decomposition. The vibrations of the compounded
molecule as a whole might cause the visibility of the
fundamental sensation of light. The " Purkinje effect "
has been described at p. 146. It must be pointed out
that a similar effect is found in a photographic plate
which is rendered sensitive to the whole spectrum.
Such a plate, when exposed to a fairly bright spectrum,
can be caused to show a negative in which every colour
will give the same density of deposit. If everything
remains the same, except that the brightness of the
spectrum is very much diminished (but though the time

222 RESEARCHES IN COLOUR VISION

of exposure is prolonged to meet ihe dimmished biight-
ueas), the resulting negative ^ will show the red as having
very little density compared with the blue and the
green. Should the visual sensations be primarily due to
the chemical decomposition of some substance on the
retina, it would not be unexpected, if the retina exhibited
the same characteristics as found in the photographic
plate. This is one form which might account for the
visualisation of the three sensations, but, as said before,
it is only a guess, and we must leave it to the physi-
ologist to give a lead. Coming to the facts which give
evidence of the truth of the three-sensation theory, we
can mention one : that knowing the amount of stimula-
tion of the sensations which is given by any two spectrum
rays, we at once can tell the colour and the luminosity
of the colour which they will give by mixture in any
proportions. As we proceed to consider the phenomena
exliibited in colour vision, circumstantial evidence of the
truth of the theory will be offered from its power of
explaining them in the simplest of manners. There are
extant theories that accoimt for the different phenomena
exhibited by colour vision on a psychological basis
which at once removes them from the '* ken " of any
exact science. There is also one theory amongst others
which postulates more than three single sensations.
This must stand or fall on the evidence afforded by
observations, amongst them being those which are
recorded in this work.

> See Paper No. 29.

CHAPTER XV

THE COLOUR SENSATIONS

We can commence the practical demonstration of
the trichromatic theory of colour vision with a
reference to Clerk Maxwell's observations.

Clerk MaxxveUs Colour Apparatus.

The instrument he employed is shown in Fig. 79.
The apparatus^ really was a spectroscope, somewhat

Cc

V

Fig. 79.

the same as the colour patch apparatus described in
Chapter IV., but the paths of the rays are reversed in
the way in which it was used. In a screen at D (see
Fig. 79), three slits (X, Y, and Z) were placed, which
were viewed from the position which the collimator slit
occupies. One slit was placed in the position that a
red would occupy in the spectrum if light were sent

^ The second instrument he employed was based on the first one, which
we describe.

2SS

224 RESEARCHES IN COLOUR VISION

through the collimator, another in the green, and the
third in the blue. If the three slits were illuminated
by diffused light and the eye were placed at what would
be the collimator slit E — when the slit Y in the green
was alone opened — looking through L it would see the
surface of the prism P illuminated with a spectrum
green ; if the red or the blue were only open, then
the prism would appear illuminated by red or blue.
When all three slits were open, the colour seen would
be a mixture of all three rays. Clerk Maxwell caused
a white card, on which sunlight fell, to illuminate the
slits. A comparison white light was also furnished by
a light from a sunlit card passing between B and C,
but which did not pass through the prisms, but was
reflected by a mirror M. This white light was seen as
a square patch alongside the illuminated prisms. The
colour seen in the prisms of course depended on the
position of the slit or slits which were open.

With this apparatus Maxwell made his observations.
In the first instance, the three slits were placed in posi
tions which he selected empyrically as standard ones.
One slit illuminated the prism with a " good " red
when viewed from the eye aperture, another with a
*'good" green, and the third with a "good" blue.
The slits were then opened or closed until the prism
was illuminated with a white which matched the
" comparison " white in hue and brightness. He next
kept two of the slits in the standard positions and
moved the third into different parts of the spectrum,
and again matched the white as before. This slit
was then replaced in the standard position and one
of the other slits was moved in the spectrum, and again
matches of white made. Finally, the slit which had
not been moved was moved, the other two being in

mF-i-

THE COLOUR SENSATIONS 225

the standard positions, and matches once more made.
From these observations equations were formed that
included the position of the slits and its measured
aperture.

MaxweWs Colour Equations.

The following table contains the means of four sets
of observations by an observer, Clerk Maxwell, called
K, and is typical of his mode of procedure : —

Table XXXV.

44-3(20) + 31 -0(44) + 27 -7(68) x= W
16-1(28) +25-6(44) + 30-6(68) = W
220(32) 4- 12-1(44)+ 30-6(68) = W
6-4(24) + 25-2(36) + 3 1 -3(68) = W
16-3(24) + 26-0(40) + 307(68) = W
19-8(24)+35-0(46)+30-2(68)=W
21-2(24) + 41-4(48)+27-0(68) = W
22-0(24) + 62 0(52) + 1 3-0(68) = W
21-7(24) + 10-4(44) + 61-7(56) = W
20-5(24) + 23-7(44) +40-5(60) = W
19 7(24) +30-3(44) +33-7(64) =W
18-0(24)+31-2(44)+32-3(72) = W
17-5(24) + 307(44) +44-0(76) « W
1 8-3(24) + 33 -2(44) + 63-7(80) = W

(The figures in brackets indicate the place in the
spectrum the slits occupied. W is white, always of
the same value, which was matched by the mixed
colours. )

These equations were referred to the standard equa-
tion, which was the mean of twenty observations with
the slits at the standard places (24), (44), and (68) —

18-6(24) + 31-4(44) + 30-5(68) = W

Incidentally Maxwell remarked that from these

twenty equations the mean error of the red was -54,

p

226 RESEARCHES IN COLOUR VISION

of the green 1*22, and of the blue ri5, whilst the
error of mixture of R, G» and B to make white was
2 '67. The mean error in differences of the amount of
two colours in a mixture is only about '85, and as the
hue of the mixture depends on the ratios of the com-
ponents, whilst the brightness (luminosity) depends
upon their sum, it appeared to him that the eye \& a
more accurate judge of the identity of colour in two
parts of the field of view than of their equal illumina-
tion.

By eliminating W from the fourteen equations in
the table by means of the standard equation, the
different rays of the spectrum are shown in terms of
the three standard colours he selected, and are as
follows : —

Table XXXVI.

(24) (44) (68)
44-3(20) « 18-6-f- 0-4+ 2-8
161(28)- 18-6-h 6-8- 01
22-0(32) « 18«+19-3- 01
25-2(36)= 12-2+31-4- 08
26-0(40)= 3-3 +31 -4- 0-2

350(46)= -1-2 +31 -4+ 0-3
41-4(48) = -2-6+31-4+ 3-5
62-0(52) = -3-4+31-4+ 176
61-7(66)« -3-1+21 -0+30-6
40-6(60)= -1-9+ 7-7 +30-6

33-7(64) = -11+ 11+306

32-3(72) = + -6+ 0-2+ 30-6
440(76) = + M+ 0-7+30-6
63-7(80) = +0-3- 1-8+30-6

(The three standard colours are of course omitted, as
they would be equated to themselves.)

The figure shows the results of the equations dia-
grammatically as given by MaxwelL

THE COLOUR SENSATIONS 227

It will be noticed that there are parts of the three
curves below the base line. These are the negative
quantities in the equations after the left-hand members
have been reduced to unity. We shall find that these
negative quantities are due to the fact that most of the
spectrum colours contain an appreciable quantity of

white. If this white were deducted from the white
which the colours matched, the negative values would be
non-existent. (The addition of the ordinates to one
another to make a luminosity curve is rather misleading,
as it is only the widths of slits and not the luminosities
which are added together.)

Maxwell's Slit Apertures turned into Luminosities.

An attempt to turn these measures into a true
luminosity curve has been made by using the lumi-
nosities of Maxwell's three standard rays, as found in
the solar spectrum of a mid-day sun in June and also

228 RESEARCHES IN COLOUR VISION

in October, and an example is given. That the two
curves differ is not surprising when the table is scru-
tinised. The width of the slit through which the red
rays pass is the same for the first three numbers.
From (36) to (52) the green rays have also the same
apertures of slit, as also have the blue rays from (56)
to (80). We shall see that the proportions are not
quite the same if they are compared with the sensation
curves given later on.

The equations are made so that the left-hand
member is unity, and the right-hand members arelmulti-
plied by the following luminosities of the three standard
colours : —

(24) ifl SSN. 56-3 having a luminosity of 40-3
(44) „ 40 „ „ 63

(68) „ 20-3 „ „ 2-6

Table XXXVII.

Maxwell's
Scale.

(20)
(23)
(28)

(32)
(36)
(40)

(44)
(46)
(48)

(52)
(56)
(60)

(64)

(72)
(76)
(80)

Measured

SSN. •

• Laminodty.

Solar Spectrum

Luminosity.

69-6

18

•3

66-3

40-3

40*3

63

70

76

49*8

90

99

46-6

98

94*6

43-2

81*2

81

40

63

.02^

38-3

62

64

36-7

46

41-6

33-4

30-4

21

30-1

20-6

11-6

26-9

119

6-8

23-6

2'i

4-4

20-3
IM

3-3

1-7

13-8

3-7

1-2

10-6 , .

• ' -'* .

•6

BBS

THE COLOUK SENSATIONS 229

Colour Sensations.

We will now proceed to describe the method by which
-the writer worked out his own colour sensations. It
may here be stated that the writer's colour sense is
normal, as is also his form vision.

First of all, let us place a slit in the red near the
extreme end of one spectrum, and in a second spectrum,
as formed by the apparatus described at p. 44, let us
place another slit in a movable slide, so that it can be
put in any part of the second spectrum desired ; and let
us place the two patches side by side. Let a sector be
placed in the path of the second spectrum's rays. If we
place the slit at SSN. 58, which is a red, and equalise the
luminosity of the two patches, we shall find a slight
difference in hue. If we move the slit to 60, we shall
find that the hues of the two patches are the same. Such
is the case at SSN.'s 61, 62, and 63, the first slit being at
SSN. 64 (with a piece of red glass in fi-ont of the slit to
destroy the effect of any stray light which may be about).
Thus we may take it that a slit placed anywhere from a
little above SSN.'s 60 to 64 will show the same hue, and
this includes the place which the red line of the vapour
of lithium occupies when the salt is volatilised in the arc
or is heated in a spirit or gas flame. As regards the
violet of the spectrum, it similarly appears to be of
one uniform tint throughout when the necessary pre-
cautions are taken to prevent its contamination by any
white light which may come from the illumination of
the prisms. If we take a slice of violet light from
the spectrum and form a patch with it from one
spectrum, and mix a very minute portion of white
light with it, we shall find that it becomes lavender
coloured. When therefore repeating the experiments

230 RESEARCHES IN COLOUR VISION

made at the red end in the violet, it is well to place in
front of the slit a piece of cobalt blue glass or ammonia
sulphate of copper. This cuts off the green, and very
nearly all the yellow and red, but allows the violet to
pass, so that any contamination is a minimum, all the
brightest parts of the spectrum being cut off from any
small quantity of white light which may struggle
through the slit. When this precaution is taken, it
is found that the hue of the region from near G in
the solar spectrum upwards is the same, the only
difference being its brightness at the different parts.
But violet is not a primary colour, for if we take a
patch of violet light from one spectrum and place one
slit in the red near the red lithium line, and another
in the blue near the blue lithium line, we can make a
mixture of red and blue which will match the violet,
to which a little white has been added. We shall see
hereafter that the blue itself contains a large percentage
of white, and for this reason white has to be added to
the violet. This tells us that as we require pure colours
— i.e. unmixed with white as far as possible — for making
colour mixtures, it is as well to use violet as one colour
(remembering that it is a definite mixture of red and
blue), in preference to the blue, which is contaminated
by inherent white. When a mixture is made, the violet
can always be converted into blue and red, and the latter
be added to the red which may be in the mixture.

Colours not ideriticcd with Colour Sensations.

So far we have been dealing only with colours, and
not with colour sensations. If Helmholtz's diagram,
p. 214, were correct, one colour would never stimulate
one sensation by itself. As it is, however, the red

THE COLOUR SENSATIONS 231

stimulates only the red sensation in one part of the
spectrum, whilst the violet stimulates both the red and
blue, and not the green sensations. A green colour
not only stimulates the green sensation, but it stimu-
lates the red and blue sensations as well, as is shown
in Helmholtz's diagram. The trichromatic sensation
theory requires this to be the case.

That there is white mixed with the purest green, we
shall demonstrate experimentally. Now, white involves
the stimulation of all three sensations, so that no gi*een
can represent the pure green sensation to the normal
vision, though, as we shall see, it can be felt by one form
of colour blindness. White is the only mixture with a
green sensation which can help us to realise most nearly
the kind of sensation that it is, and one of the first
searches to be made is to find some colour in which this
is the only admixture.

Equal Stimulations of the Three Sensations to

produce White}

It is well, as a preliminary, to consider the sensation
of white as the result of the equal stimulation of the
red, green, and violet perceiving apparatus, remembering,
of course, that the violet is compounded in definite pro-
portions of red and blue sensations. We may use it
as a temporary sensation without objection if this
be remembered.

We can then construct diagrams which will show
what points in the spectrum can be fixed by preliminary
observations.

A, B, and C are the most interesting cases. Let the
stimulation of the sensations be represented by vertical
lines. In A we have the red and green sensations of

> See Papers Nos. 5 and 6.

232 RESEARCHES IN COLOUR VISION

/

i

equal heights, but V is less. Drawing a horizontal line
through c, aR, 6G, and cV, represent equal stimulations
which make white, leaving da and eh equal. We thus
have a colour which is made up of a mixture of R and G
sensations (RS. and GS.), together with white. Now,
equal stimulation of RS. and GS., we shall see later, give
the sensation of yellow. If we place a slit in the violet
and move it along the less refrangible part of the
spectrum, we shall find a place where this colour and
violet together make a white (the slits are opened or

a^

^

R G V
A

G

B

G V

Fig. 81

closed to make the match). This position, then, i? that
in which the red and green sensations are equally stimu-
lated, and answers to A. In B we have a green ant
violet with equal ordinates and a deficiency of red. If
we place a slit in the red and move another about in
the green, we shall find a colour which with the red
makes white. This position, then, will have an equal
stimulation of green and violet. This gives another
fixed point. The next point to determine is diagram-
matically shown by C, which illustrates the green we
have to look for, mixed only with white. This is more
difficult to find, as it would require a purple to be
added to make a match with the white, and this does
not exist in the spectrum. Suppose we mix A with B,
we get a diagram of the kind shown in the fourth dia-

i

THE COLOUR SENSATIONS 233

gram. There are equal reds and violets stimulated,
but a larger stimulation of green sensation. This gives
a colour paler than the spectrum colour, but still a
green which can be matched. There are also other
plans dependent on trial and error of fixing this point
which can be carried out. (There is also a confirmation
which can be made by a green-blind person, of which
we shall speak presently.) At any rate, we have several
data with which to commence a series of observations.

Conditions to he observed in making Measures.

There are several considerations that have to be taken
into account in making measures. In the first place,
the white light used must be of the same " quality " —
that is, the relative luminosities of the different rays
of its spectrum must be constant. Secondly, the
measures are best made with the central part of the
retina, and the patches of mixed light should be "^
the same size and be viewed from the same distance -
throughout. The light firom the crater of the positive
pole of the arc light is always of the same quality, and is
best adapted for a standard light when colour patches
have to be viewed ; and in fixing the points in the
spectrum the above conditions should be carefully
carried out.

When the observations^ for obtaining the fixed
points have been made, it will be found that the colour
which with violet makes white is at SSN. 487, that
the colour which with red makes white is at SSN. 34*6,
and that where the green, mixed only with white is
found, will be SSN. 375. With these three points fixed,

^ The following observationa were made with the spectrum of the crater
of the arc light with sloping carbons.

234 RESEARCHES IN COLOUR VISION

and the knowledge that the red stimulates a pure sen-
sation of red, and the violet sensations of blue and red
unmixed with green, we can begin to find the sensations
which exist in other colours. What first is required is
to know the amount of white which exists in the green
at SSN. 37*5. To ascertain this we must place one slit
at SSN. 37-5 and another at, say, SSN. 59'8, the position
of the red lithium line. The luminosities of these two
colours with equally wide slits must be taken, say,
against a neutral colour, such as yellow or white. They
will be found to be 39*2 and 9 4 respectively. A patch
of orange light from the second spectrum is placed
alongside the patch of mixed red and green, and an
endeavour roust be made to get the same hve of orange-
yellow in the mixture. It will be found that the mixed
lights are always paler than the spectrum colour. White
light is next added to the latter until the same paleness
of hue is produced.

The widths of the slits are measured and the lumi-
nosity of the white added is determined. An equation
is formed in luminosities thus —

a (yellow) + b (white) = c (red) + d (green)

Now, the red contains no white, so all the white that is
in the mixture must be white contained in the green
colour. The equation comes out —

' a (yellow) =* c (red) + [d (green) - h (white)]
That is, the percentage of white in the green is -^ x 100,

and the percentage of green sensation is ^ — -s— x 100.

When these equations have been worked out, it will
be found that the white (obtained from the mean of

■ II i—— ^1^1^^— i^l^^— Pl^P— ^I^P^^^^^^^^^^ "W^^F" T'"l '^m ■ "- ~ -.agj^^^—— _-«.«^— llJiii

THE COLOUR SENSATIONS 235

several equations) in SSN. 37*5 is 69 per cent, of the
luminosity. The following is a concrete example of the
observations made. A yellow was taken at SSN. 50'05,
and the following equation in luminosities obtained : —

RS. (37-6). (6005). WTiite.

487+45-8 = 63 + 31-5

As there is no white in the red sensation (RS.), it follows
that the 31 5 white is in the SSN. 37-5. This gives —

(6005). RS. GS.

63 = 48-7 + 14-3

That is to say, from this equation there is 31*2 per cent,
of GS. in 37-5 and the white in SSN. 37*5 is 68-8 per
cent.

The composition of the orange and yellow regions of
the spectrum was found by placing one slit in the red of
the spectrum and another in the yellow or at D, the
composition of these rays having been determined by
the observations which were made to find the percentage
of white in SSN. 37*5. Some dozen colours between
SSN.'s 49 and 58 were determined in this way, bearing
in mind the small corrections due to the shift in hue by
the addition of white.

When once this percentage of white in the green
has been arrived at, the percentage sensation composition
in luminosities of the remaining colours can be readily
found.

By putting three slits in the spectrum and fixing
one in the violet about SSN. 10 and another at SSN. 59-8,
and putting the third slit at diflferent positions between
SSN. 35-5 and SSN. 48*7, equations can be formed of the
luminosities of the three colours necessary to match the
white patch. Instead of altering the width of the slits

236 RESEARCHES IN COLOUR VISION

to make the luminosity the same as the white patch,
sectors can be put in the path of the white beam and
the luminosity of the white determined. The standard
equation, to which all other equations are referred^ is the
equation given by placing the " green " slit at SSN. 37*5,
the othei' two remaining as above. Thus an equation
of this form is found —

Red. Green. Violet. White.

a '\' h -V c =^ d

We have to deduct 69 per cent, of white from the gi'een
on one side of the equation and the same amount from
the other, which will give the white in terms of sensa-
tions only. It was found that the mean of the equations
gave the following as the value of white in sensation
luminosities (RS. and GS. standing for red and green
sensations and V. for violet) : —

RS. GS. V. White.

68-4 + 30-2 + 1-4 = 100

To this standard equation all other equations were
equated. The following is an example of an observa-
tion, and the calculations by which the percentage com-
position in sensation luminosity of the ray in question
was found. The ray whose composition was required to
be found was SSN. 40. It was found that when the
matched white was 100, the following was the equa-
tion, the apertures of the slits being multiplied by the
luminosities : —

RS. (40). V. White.

36-8 + 62-l + l-14 = 100
but—

RS. GS. V. White.

68-4 + 30-2 + l-4 = 100
this being the standard equation.

THE COLOUR SENSATIONS 237

From these we get —

(40). RS. GS. V.

100 = 51 + 48-6 + '42

which is the percentage composition of SSN. 40. The
composition of other rays between the red and SSN. 37*5
was found in the same way.

It was believed (ui\til the change in hue caused by
the addition of white to a colour was determined) that
there was a plan by which the amount of violet in the
SSN.'s from 37*5 to the red could be better determined
than by the ordinary equations. The idea was to
accurately determine the red to the green sensations
by this last plan, and then to mix a red at the red
lithium line with a green at 37*5 to match the hue of
the colours within that region. The white contained
in the green was known, and prima fcLcie it was supposed
that the violet necessary to form the white would be
a correct measure of the violet to be found in the ray
under consideration. The violet was in this way found,
with the result that the sensation, instead of gradually
diminishing towards SSN. 50, rose in the middle, and
had a maximum about SSN. 42. This method evidently
is inaccurate in consequence of the change in hue. For
this reason the older method has had again to be resorted
to and the violet determined from the mean of several
separate equations.

From SSN. 36 to 12 a different method was adopted,
which gave very accurate results. The composition of
all rays from SSN. 64 to 37 '5 is known from previous
observation. If, then, we place a slit at some place
having a lower SSN. than 36, we can find some colour
which, when mixed with it, will give a white. (It is
convenient in this observation to use the two spectra

238 RESEARCHES IN COLOUR VISION

given by the double colour patch apparatus described at
p. 44.) The colour being found which makes the match,
the slits are measured as usual, and the luminosity being
known, a luminosity equation is formed. Take an
example of SSN. 25*5. It was found that the ray at
SSN. 49*05 gave the white. After converting the width
of slits into luminosities, and reducing the equation so
that white was 100, the following was obtained : —

(49-05). (25-6). White.

96 + 40 = 100
Now, from the percentages already determined —

RS. GS. (49-05).

70-1 +29-9 = 100
after converting (49'05) 9(} into RS. and GS., we get —

RS. GS. (25-5.)

67-30 + 28-10 + 4-00 =100
Equating this with the standard equation, we get —

25-5. RS. GS. v.

100 = 27-5 + 37-5 + 35

In this manner the composition of all the SSN.'s smaller
than SSN. 36 were calculated.

The method of finding the percentage sensations
existing in each colour of the spectrum has now been
shown, and from the determinations ciurves of red and
green sensations and of violet were drawn as smooth
curves. The ordinates of these curves are given in the
following table in, columns IV., V., and VI. Columns
I., II., and III. are the scale numbers, the wave-lengths,
and the luminosity of the rays for normal vision.

Columns VII., VIII., and IX. give the luminosities
of the two sensations and violet, obtained by multiplying

THE COLOUR SENSATIONS

239

•061- X sa

•

N-^

>

•fZx

'SO

m

>

03

X

•

PQ

>

'S

c

mm*

s

^

t-H

1— t

S

m

1-4

•

o

00

X

••A
o

n

*-•

8

m

X

a

5

C

O

•

X

S

o

p*

OS

i : eoiotooow ^^SS^S ^S^^S SSSSSS S'^'^^^^

?pf9^ «oo^$§;S S:2$S?S SSS§^$|^ li

o
a

a

c
o

to
O

S

6
I

d

I

&

CO

CO

O

•

A^nonimmj

•

1-4

•OK 9l«0g

odb-eseiQQ ooi-i<^o^ ^909<it«^ ^S'^%9 m«^£3P^

: : : : : ^wegs'Z ooow'SPt* q^5p-i*« c>aD»o«5 ^!&S2SS?

. . . : : : ©opoo o,H^FHrH c^ci$4C4cl f5iHi-i»-i© poopo

SSfS 9&S?S S?$^? p^s^s ?9

lA

bo^.^^ rHGM-^CO iOPC4l3^ h.'^t-HOdfc^ COtOeOCOCNl r-l tH r-i O Q
C<lt>- Ot<-e<lQlO lA^CQ^lQ COiHCOCOeq r-liHi-4

• • •

• • » •

zz^

»^«S^c5 S^S585?5 85aS88

•• ^ '^^^Sa §5fe:S5^ SSggS g^JSS* ^*^ '

ipipoi pp^^oo ipp^*H^ ppiSS^ 00

SpQ o^^P^iO t«e9b«ibw a6cco>e>4C9 'rfpooioo rnciie^ciei c<icqcic9C4
©o oSSwoOb- eeosioiSio '3»^^'^"* ■^JoZeet* ir^t-t-t^t- t*t-t*b-i>-.

ootAc^©© p©c^co© pQOtoe^co ©
: : : : : : : : ©©i-Hr-ic^ o^oS-^afip t<-ooaSS>ao b-

©o© ©©©©a
_ Q 2 lO cpoo9©tf

N»5"*»0iQS t^6Bc6o6oO t^p25^55 «i-if-i.HC

^a$^ S^^$S^ ^|3^$ ^i^SSS S3
^ ^ = ' ^^28 53888S SSS^^« ^^

oit* ©t-M©»o - - -

©t-e^©^ ^9900^10 t«©ioeoe4 th

• •

• ■ ^ • • ■ ■

• • • ■ • • •

-^^'^^ii 2553^8^ S8SS8 88888

ipapqp pfHp«0 qpioSfHS 2Sm CO «o lO

: : ^i^oigg ggg^;^ S2SS3 S^SS* ^'^' =

So tH ©

888 8«8SS^8feSi3§!g5JfeS8^S85Si^'^'^ '• '•

lO O e<IOt« OO^-^rH SS5S^^ 82SSS8

^ 5^3888 SJfcSSg S^SIS**^ ^c^^r^^

M

b-<o<b <o<ocoo2S lOoAioio iom

t>-t»^lOlQ 9^«^<NC^ iHr-if->©©
^^^^9 ^^^^^ ^■^•V?^

S:S8SS

s@8 sssss 9^999 ssii^^^s i^s^^g^s ss;!:^s

© Goco^ei©

240 RESEARCHES IN COLOUR VISION

the percentages by the luminosities in column III., and
dividing by 100. A series of observations made on the
composition of the violet, show that it contains cloeely
72 per cent, of red sensation and 28 per cent, only of
blue.

Columns X., XI., and XII. show the percentage

. composition in terms of the red (RS.), green (GS.), and

blue (BS.) sensations. Column X. is column VII., to

which 72 per cent, of the violet percentage has been

added, and column XII. is 28 per cent, of the violet.

Fio. BZ. — Percentage oompoiItioD of the spectram ooloura Id leminoiities
of red, green, luid blae aetiEalioDg.

Columns XIII., XIV., and XV. show the luminosities in
RS., GS., and BS., and Columns XVI. and XVII. show
columns XIV. and XV. multiplied by 2-3 and 190 re-
spectively. Theee multipliers make the areas of all the
three luminosity curves equal. Thus, columns XIII.,
XVI., and XVII. give the stimulations, when equal
stimulations are supposed to give a sensation of white.
We shall ^ee that this is of some importance. Figure
83 shows the curves' of equal stimulation and Fig. (82)

' Tba peneDtige compoDoata of the aensationa in terms o( «qual stimu-
ktioD wiU b» foaiid m Chapter XXV.

THE COLOUR SENSATIONS 241

the percentage curve in luminosities of the three senea-
tions. At SSN. 48-6 we have the red and green curves
cutting one another, which is one of the points we have
already found. At 34'5 we have the intersection of the
green and blue curves with the red curve below, and
this is the point which with added red makes white, also
previously determined.

Fia. 83. — Seusatiou cnrres having equal area« (equal ordinatea at vaj point
make white].

At SSN. 37"5 we have the place where the green
sensation is unmixed with anything except white, a
position we have also previously determined. At this
point the red and blue curves cut, and the green curve
is here above the other two, showing that there is white
and a surplus of green.

In Fig- 84 is shown the three sensations in terms of
luminosity when the white has been deducted from
them. This is drawn from Table XXXIX.

Q

242 RESEARCHES IN COLOUR VISION

Table XXXIX.

I.

II.

III.

IV.

V.

VI,

VII.

VIII.

BSN.

Lomuiosity of senaaUon together with
the white.

Percentage compomtion of the

BensaUone, white being

deducted.

RS.
'2

OS.

• a •

BS.

W.

RS.

G&

• • «

BS.

64

■ • •

100

• • ■

62

2

■ « •

• • •

100

• • •

• « •

60

7

• • a

• A •

100

• • •

• • •

58

20-79

•21

• « ■

99

1

• • ■

56

4775

2-25

• • •

96-5

4-5

• • •

54

72-4

7-6

• • ■

90^5

9^5

• • «

52

80-64

1536

• • •

84*2

15^8

• « •

60

75

25

« • •

76

25

• • •

48

61-4

30-3

5-3

66-9

331

* • •

46

54-1

30-8

71

63-7

36-3

• • •

44

37-2

28-8

9

66'4

43-6

• • *

42

26-4

24-9

ir2

48-6

61-4

• • ■

40

14-2

19-3

16-5

42^4

57-6

• « •

38

•7

15-2

20-1

45

966

■ ■ •

36

• « •

8

1)31

16

■ • •

99-6

•386

34

• • •

51

•088

9

• • ■

98-31

1-69

32

• ■ •

3-2

•125

52

• • •

96*24

3-76

30

« ■ •

212

•155

3-43

• * «

93-18

6-82

28

• • ■

133

•192

2-48

• • •

87-37

12-63

26

■ • ■

•53

•235

2-03

• • •

68-8

31-6

24

• • •

•03

•26

1-66

« • •

10-5

89-5

22

•43

• • ■

•246

•73

15

• • •

86

20

•64

• • •

•235

•33

69-7

• • •

303

18

•61

• • •

•201

•15

71-8

• « •

28-2

16

•49

• • •

•18

•03

731

■ * •

26-9

14

•39

• • •

•154

• ■ •

72

• • ■

28

12

•334

• • ■

•126

« ■ •

72

• • ■

28

10

•253

• ■ •

•098

• ■ •

72

• • •

28

8

•187

« ■ •

•073

• • •

72

• • •

28

6

•13

• • •

•051

• • •

72

• • •

28

4

•101

• • «

•039

■ • •

72

• • ■

28

2

•072

• • •

•028

• • •

72

• « «

28

Areas.

'057

• • ■

•022
2-53

■ • •

72

■ • •

28

450

192

187

This table is useful to have by one, as it simplifies
the calculation for obtaining the true dominant colours
of pigments, &c. From columns XIII., XVL, and

THE COLOUR SENSATIONS 243

XVII. of Table XXXVIII. the columns in this table
are readily found.

FlO. 84. — Luminosil; carves of red, green, bine, and whits sensations ot tbe
prismatic spectrum ot tfae crater (positive pole) of the arc light.

[Taking SSN. 40, for instance, we find that SSN. 40
has —

Col. xm. Col. XVI. Col. XVII.

RS. GS. B8.

25-61 55-89 114

As equal ordinates make white, the smallest ordinate,
11'4 in this case, must be deducted from the other
two to obtain the white, and we have left —

Ra GS. X 2 3.

14-2 and 445

Thus, after deducting 16 -5 of white, the amounts of RS.

and GS. are 14'2 and -oTo" , or 193, and the colour is

denoted by the equation—

KS. GS. W.

14-2 + 19-3 + 16-5 = 50

In the same way the equations for the other colours
are calculated, and we have from the results Table
XXXIX. and Fig. 84.]

244 RESEARCHES IN COLOUR VISION

The following tables show the luminosity composition
— (1) of the arc light with horizontal positive pole;
(2) of the Nemst lamp : —

Table XL. — Sensaium Luminosities^ and Etjual Stimidcdion OrdincUes
of an Are Light toith Horizontal Positive Pole,

SSN.

7217

Lmninoflity.
•5

RS.

oa

BS.

• • •

OS. X 2-21.

« ■ •

Baxll7.

64

•5

• * ■

• • •

62

6957

2

2

• ■ •

• ■ »

■ • •

• • *

60

6728

8-7

8-7

• • •

• • •

• • •

• • ■

58

6521

21-5

21-3

•2

• • ■

•44

• • •

56

6330

48-3

461

2-2

• « •

4-86

• « •

54

6152

70

63-3

67

• • •

148

• • •

52

5996

84-7

71-3

134

■ ■ •

29*61

• ■ •

50

5850

96-2

721

241

• • ■

53-26

• • •

48

5720

99-9

67

329

•02

72-7

2-33

46

5596

95

59

36

•029

795

3-39

44

5481

85-3

49-7

366

•036

78^7

41

42

5373

72

39-6

32-3

•048

714

6-61

40

6270

561

291

27-

•076

59-8

8-89

38

5172

41

199

21

•094

463

11

36

5085

27-5

12-7

14-8

•11

32-5

12-8

34

5002

15-8

6-9

8^7

•127

193

148

32

4924

8*9

3-75

5

•151

11-1

17-6

30

4848

617

216

337

•195

7-45

22-8

28

4776

4-6

206

2-31

•247

5-1

28-9

26

4707

3-5

177

1-44

•307

318

36-9

24

4639

2-7

1-58

•76

•363

168

425

22

4578

216

1-4

•36

•381

•78

44-6

20

4517

.1-76

1-24

■14

•378

•31

44-2

18

4459

1-48

107

•07

•36

•15

421

16

4404

1-29

•93

•03

•332

•06

38-9

14

4349

11

•79

•005

•302

•01

353

12

4296

1

•7

• • •

•27

• • •

316

10

4245

•85

•61

■ • •

•288

• • •

27^8

8

4198

•73

•63

• « ■

■204

« ••

23-8

6

4151

•62

•45

• • ■

•174

• « ■

20-3

4

4106

•5

•36

• ■ •

•14

■ • •

164

2

4062

•4

•29

• • •

•112

• • «

131

4010

•3

•22

• • •

•084

t ■ •

9-8

^ The percentage composition of the SSN.'s is the same as that given in
Table XXXVIII.

THE COLOUK SENSATIONS

245

Table XLI. — Sensation Luminotities ' ami FajhoI Stimulation
Ordinates of a Nemst Lamp Light. 1 amp. 100 vols.

SSN.

7217

Luminoflity.

1

OS.

BS.

OS. X 2-77.

• • •

BS.X264.

64,

1

1

t

••• 1

• • •

• • •

62

6967

6

5

• • >

• ■ •

...

60

6728

12

12

1

■ • •

• * ■

ft • •

68

6521

31-5

311

•4

11

• • *

66

6330

66

621

2-9

8

• ■ •

64

6162

87^5

80-6

7

• ■ •

19-4

• > •

62

6996

99-7

83-9

15-8

• • •

437

• • •

60

6860

98^8

741

24-7 i

• • ■

68-4

• ■ •

48

6720

89

69-7

29-4

•0178

814

452

46

6696

76-6

47-7

29

•0237

803

602

44

6481

61-6

367

25^8

•0258

714

6-56

42

6373

46-7

25

21

•0313

582

796

40

5270

35

18

17

•0410

471

1041

38

6172

24-5

11-9

12-6

•0563

34-9

143

36

6086

16

6-9

81

•0630

224

16

34

5002

7-6

325

4-25

•0652

118

16-52

32

4924

4

166

2-24

•0680

6-2

1727

30

4848

2-5

1-2

1-35

•0790

3-74

2006

28

4776

1-8

•8

•91

•0936

252

2377

26

4707

1-6

•76

•61

•1302

17

3306

24

4639

1-25

•72

•36

•1680

•97

4267

22

4678

106

•69

•16

•1862

•44

47^04

20

4617

•9

•64

•07

1

•1917

•19

48^68

18

4469

•76

•64

1

' 03

•1262

■09

44-75

16

4404

•6

•43

• ■ •

•1660

■ • •

3962

14

4349

•6

•36

• ■ •

•1400

• ■ •

3656

12

4296

•4

•28

• ■ •

•1120

• • •

28^44

10

4246

•35

•26

i

• • •

•0980

• • •

2489

8

4198

•3

•22

■ • •

i 0840

• ■ ■

2134

6

4161

•25

•18

■ • •

•0700

• • ■

1778

4

4106

•21

•16

• ■ •

•0588

• • ■

16

2

4062

•18

•13

• • •

•0524

■ • •

13-31

4010

•15

•11

• ■ ■

•0420

• • •

10-67

It may be useful for reference to have the sensation
luminosities of the normal spectrum in which the abscissae

' The percentage composition of the SSN.'s is the same as that given in
Table XXXVIH.

246 RESEARCHES IN COLOUR VISION

nal FpectTnm) carrcs of equally nt.imulated red, green, and blot
at to form the wfalte of the arc light nith sloping carbonii.

Table XLIL— Normal Speetrt

apHtram

6800

notily.

E^

aa

«.

K8.

08.

B8.

OS.XfK.

BS.X16S.

100

1

6700

6

100

6000

10

09-7

■3

9-97

■03

■07

6S00

17

98-6

1-5

16-74

-26

■6

6400

26

97

■3

26-22

■78

1-9

6300

41

96

■6

38-96

2-05

4^9

6200

6fl

02

8

54-28

4-72

11-2

6100

75

88-G

11-5

66 25

8-76

20-8

6000

85

84

16

714

13-6

32-4

6900

93

78-5

21-5

7293

20^

48-7

fiSOO

99

72

28

71-28

27-72

66

6700

100

657

34-3

■022

66-7

343

■022

81-6

3-4

5600

95

62

38-6

■031

68-9

36-6

■029

87-1

4-5

6600

89

58-5

41-5

•041

52-06

38-93

■036

87-9

6-6

6400

80

55-5

44'6

■058

44-4

36-6

■046

84-7

7-1

5300

70

52'3

47-2

■1

3661

3304

-07

78 6

10-9

5200

54

49-3

60-6

■185

26-62

27-27

-1

641

16-5

6100

30

46-5

531

■4

13-95

15-93

■12

37-8

18^6

5000

18

43-8

65-3

■86

788

9-95

■155

23-7

24

4900

11

42

66

2

616

■22

14-7

34-1

4800

7-6

43

82-4

4-6

323

3^93

■346

9 4

33-5

4700

5

60

41-3

8^7

2-6

2-06

-436

49

67-4

4600

3-6

62

21-8

162

2-17

-76

■567

1-8

82-9

4600

2-7

72

7-3

2b7

1-94

-2

■586

■6

90-8

4400

21

72

2-2

258

1-61

■05

-642

-1

86-6

4300

17

72

28

1^22

■476

75-8

4200

1-3

72

28

■94

■367

56-9

4100

1

72

■72

■28

434

4000

■75

72

■64

■21

32-5

3900

■5

72

28

■27

•U

21-7

3800

■26

72

28

•13

■07

10-8

THE COLOUK SENSATIONS 247

are wave-lengths. Tl)e luminosity curve ^ of the normal
spectrum is given ; knowing the percentage composition in
sensations of the diflPerent wave-lengths (X), the sensa-
tion curves for the normal luminosity curve is readily
obtained by multiplying the luminosity by the per-
centages and dividing by 100.

> Taken with a grating ruled on flat glass. The ruled surface was
silvered, and the silvered gurface was used to form the spectrum.

CHAPTER XVI

COLOUR SENSATIONS IN COLOUR DISCS

Having found the colour sensations for the spectrum,
we are now in a position to utilise them in the deter-
mination of the colour sensations stimulated by pigments.
In Chapter XI. we have shown how three standard
pigmented surfaces of red, green, and blue can be used
for the determination of the luminosities of other pig-
ments when the colour discs are brought into requisition.
It is possible now to extend the usefulness of the stan-
dard pigments in the directions of detecting colour
blindness quantitatively, and also for ascertaining the
proper colour screens to be employed in colour photo-
graphy. It must be premised that transparent coloured
media can be examined in exactly the same way as the
pigments are examined.

The fii st step to take is to get the percentage absorp-
tion of the coloured bodies, whether transparent or
opaque. The method of doing this has been described
in Chapter XI., and need not be repeated. Having
obtained the percentage absorption, the next step is to
connect these results with luminosities. The lumi-
nosities depend on the luminosity curve of the white
light employed to form the spectrum and with which
the colours are also illuminated.

Having found the luminosity curve of the white, the
percentage of absorption is multiplied by the luminosity
of the spectrum, and this gives the luminosity curve of
the transparent coloured body, or of the pigment. The

248

COLOUR SENSATIONS IN COLOUR DISCS 249

luminosity at any point of the curve, when multiplied
by the percentage composition of the colour at that
point, will give the sensation luminosities of the pig-
ment (or transparent body). (We can also get the
colour sensations evoked by the one or the other by
multiplying the luminosity of the different sensations
at any part of the spectrum by the percentage absorp-
tion. We then get the luminosity curves in terms of
the sensations.) By calculating from the colour equation
to the white, the proportion of the three sensations
required to form it, we can readily obtain the amount
of white which exists in the pigment (or transparent
body). This is most readily accomplished by multi-
plying the green and the blue sensation ordinates by
the factors which, in the naked spectrum, give equal
areas. The smallest of the resulting areas is deducted
from the other two, and after reconverting into ordinary
luminosities there will only be two sensations remaining
and white. This will perhaps be more easily under-
stood by working out an example. We give in detail
the measures and calculated sensation luminosities for an
emerald green, in the light of the arc (sloping carbons).
The equation to the light of the naked spectrum in
luminosities is —

RS. GS. BS.

68-4 + 31+-58

To make equal areas to show equal stimulation to
form white, we have to multiply the GS. by 2*2 and the
BS. by 117. So (see Table XLIII.) we have to multiply
141 by 2-2, which is 311-6, and 1*4154 by 117, which is
165, whilst the RS. is 211-4. This makes BS. the
smallest area. Deducting this from the (equal stimu-
lation) GS., and dividing by 2*2, we get GS. 66*3, and
white 241-7.

250 RESEARCHES IN COLOUR VISION

Tabu: XLIIL — Colour Sensatuma of Emerald Oreen in an

Are Spectrum,

SSN.

62
60

58
56
54
52
50

48
46
44
42
40

38
36
34
32
30

28
26
24
22
20

18
16
14
12
10

1

Lnminositiea.

Abfiorption
Percentage.

1

1 Laminoeitj.

.

3

KS.

GS.

BS.

•02

■02

■ • •

• • ■

5

i -05

■05

i

1

• • •

7-7

1-5

1-57

1

•015

• • «

10-5

5-2

5-02

•23

• • •

14

9-8

8-88

•92

• ■ •

20-5

15-5

13-06

2-46

• « •

24-5

23-6

17-73

5-90

• • •

33-5

33-4

22^48

10-92

•0067

43-5

41-3

25-75

15-52

•0128

57

48-7

28-38

20-35

•0204

67

49-2

26-80

22-44

'0330

72-5

40-6

21-02

19-57

•0475

76

30-8

1505

15-73

•0708

78

21-2

9-83

11-38

•0897

78

12-3

5-40

6-90

•1076

75

6-8

2-82

3-75

•1140

69

4-3

1-79

2-34

•1335

60

2^8

1-24

1-41

-1452

61

1-9

•90

-84

•1550

41-5

11

•66

-31

'1502

33-7

'7

•47

-12

•1157

25-5

•43

-31

-03

•0957

19

•29

•21

•01

•0674

11

•14

•10

• • ■

•0363

8

•03

-02

* • •

•0090

1

•01

• • •

'007

ft • •

•0027

Area

Area

Area

211-4

1410

1^4154

The equation to the emerald green thus becomes —

RS. GS. White.

45-8 + 66-3 + 2417

353

The luminosity of the pigment is ^^^ = 40*8 (865 is

865

the area of the naked spectrum luminosity).

COLOUR SENSATIONS IN COLOUR DISCS 251
The equation in percentages of this luminosity is —

RS. GS. White.

5-34 + 7-64 + 27

It may be of interest to show that the equation to
emerald green varies according to the light it is viewed in.
In the next table is given the same emerald green when
viewed in the light of a paraffin lamp. The percentage of
intensity of light after absorption is, of course, the same
as in the last case, but the luminosity is different.

Table XLIV. — Einei*ald Green in Light of a Paraffin Lamp,

SSN.

Luminosity of

Paraffin Ught

Spectrum.

Intensity.

Lnminosity.

1

RS.

•1

GS.

1

BS.

• • •

62

3-4

3

•1

• • •

60

11-3

5

•6

•56

...

• • •

58

81-3

7-7

2-6

2-38

•25

• • •

56

65

10-5

6-8

6-52

•31

• • •

54

95-7

14

13-6

12-48

1

• • •

52

100

20-5

20-5

17-26

3-24

• • •

50

89-2

24-5

21-8

16-39

54

• • •

48

69-4

33-5

23-6

15-61

7^98

•0047

46

52-7

43-5

22-8

14-26

8-53

•0070

44

39

57

21-7

12-94

8-72

•0091

42

281

67

18-8

10-46

8-34

■0126

40

20

72-5

14-5

7-47

6-96

•0170

38

13-2

76

10

4^86

5-12

•0230

36

8

78

6-2

2-89

3 28

•0257

34

4-2

78

3-2

1-37

1-81

•0281

32

2-2

75

1-63

•7

•9 :

•0277

30

1*32

69

•88

•38

•47

•0290

28

•84

60

•49

•22

•25

-0270

26

•52

51

•25

•13

•1

•0230

24

•35

41-5

•14

•08

•04

-0192

22

•22

33-7

•07

•05

•01

-0132

20 ,

•15

25-5

1

1

•02

•02
12712

1

• • •

-0080

•2743

1

696

1

62-71

252 RESEARCHES IN COLOUR VISION

From the equation to the paraffin light it is found
we have to multiply the GS. and the BS. by 3*4 and 550
respectively, to make all the areas equal. We then
have —

RS. GS. BS.

127-2 213-8 151

Here we have to deduct the red^ 127, and as a result
we get —

GS. GS. BS. BS.

?|^ = 25-3and^^ = .044
34 550

and the equation is —

GS. BS. White.

25-3 + -044 +164-7

1 QO

The luminosity is ^ x 100 = 27-3.

The equation becomes, in terms of the emerald green
luminosity, —

GS. BS. White.

4-16 + -006 + 23-14.

It will be noted that with the arc light the emerald
green reflected RS. and GS. and white, whilst here the
RS. disappears and BS. appears in its place. The lumi-
nosity also is altered with the paraflBn light ; it is 27 "3
(696, being the area of the paraffin light spectrum),
with the arc light 40-8. The reason obviously being
that the paraffin light contains very much smaller Ituni-
nosity in green rays than the arc light.

The sensation curves of two other pigments have
also been calculated out for the arc light and paraffin.
The percentage of reflection for these and some others is

COLOUE SENSATIONS IN COLOUK DISCS 253

given. If the details are required, the sensation curves
can be found by multiplying these percentages by the
sensation himinosities of the naked light.

The equation for the vermilion as viewed in the arc
light was found to be —

RS. GS. White.

142-5 + 16-5 + 5-3
or, in percentage of luminosity, —

RS. GS. White.

16-5 + 1-9 + 61

the luminosity being 245 per cent, of a white surface
illuminated by the arc light.

With the parafBn light the equation was —

RS. GS. White.

194 + 15-3 + 37
or, in percentage of luminosity, —

RS. GS. White.

39-6 + 3-1 + 7-5

the luminosity being 50 nearly.

Here we see that the extra red in the paraffin
light gives the vermilion increased luminosity.

The equation for French ultramarine blue, when
viewed in the arc (crater) light, is —

RS. BS. White.

2-32 + 1-56 + 34-2
or, in percentage of luminosity of a white surface, —

RS. BS. White.

•27 + -18 + 3-95
and it has a luminosity of 4*4 per cent, of white.

254 RESEARCHES IN COLOUR VISION
In the paraffin light it has an equation of —

GS. BS. White.

•6 + -2 + 15-64
or, in percentage of the ultramarine luminosity, —

GS. BS. White.

•012 + -004 + 3-2

as it has a luminosity of only 3*2 per cent, of a white
surface.

"^^

CHAPTER XVII

CHANGE IN HUE OF COLOURS BY THE ADDITION OF
WHITE LIGHT, AND THE AMOUNT OF COLOUR
WHICH WOULD BE ADDED TO WHITE WITHOUT
BEING PERCEIVED 1

When a spectrum is placed upon a screen and a patch
of white light is caused to travel along it, more par-
ticularly if the white light of the arc crater be confined
to one-half of the breadth of the spectrum, it will be
at once apparent that there is a change in the hue of
the colour. In the red the colour becomes pinker as
more of the white light is added, the scarlet becomes
orange, the orange yellow, and the yellow green. The
yellow-green does not suffer a change, but as the green
is approached it becomes yellower in hue, and as the
white light passes over the green, this same tendency
to yellowness appears. In the blue there is not much
alteration, but as the violet is approached a very small
quantity of white will make it appear nearly salmon-
coloured. If the white light added be that of a paraffin
lamp, the red became more orange, the scarlet, as
before, orange, the orange the colour of the white light.
Where the hue of the added white was the same as that
of the colour — that is, when the colour was nearly that
of the D light — no change took place. The behaviour
of the green was as before, as also of the blue ; the
violet became more yellow-pink than with the arc light.
This change of colour, as far as the writer knows, had
not been investigated quantitatively ; but when the

* See Paper No. 25.

265

256 KESEARCHES IN COLOUR VISION

investigation had been concluded, it explained several
phenomena which had been met with. The value of
the change of hue was ascertained in quite a simple
manner.

The apparatus employed was the double spectrum
apparatus described in Chapter IV., p. 44. In this
investigation the lowest half of the beam coming
through the prisms was deflected at right angles to
the axis of the beam by a right-angled prism, and
again deflected by a second mirror nearly parallel to its
original direction ; see p. 45.

The two spectra were of approximately the same
intensity. Two colour patches could now be formed
side by side on a white surface when slits were inserted
in each spectrum. The beam of white light reflected
from the first surface of the first prism could be thrown
on either of the patches. In the investigations here
given the white light was thrown on the right-hand
patch, which was produced by the spectrum formed from
the diverted beam.

The two colour patches overlapped each other, but
the two coloured surfaces were caused to touch each
other by inserting a rod in the path of the beams.
Another rod also cast a shadow from the white light,
so that the left-hand patch was free from any mixture of
white. Both spectra were accurately scaled, so that the
wave-lengths in both were known, and the same colour
could be placed in each patch. [The patches of light
were about 1^ ins. square.] A qualitative examination
of the patches to which white light had been added
was first undertaken. The luminosity of the white
light was made of about half the luminosity of the D
light of the diverted spectrum, so that the red and
parts of the green had, of course, a larger percentage

CHANGE IN HUE OF COLOURS 257

of white added to them than had the yellow, the yellow-
green, and the orange.

Two patches of the same red were matched in
intensity, and to the right-hand one white was gradu-
ally added. It was seen that the hue was changed,
and that the mixed light was certainly yellower than
the original colour. When the patches were orange,
the colour became decidedly yellow, and this change
in hue continued until Scale No. 487 (X 5772) was
reached, when no change in hue could be noticed.

Passing the slits into the yellow-green, the colour
lost much of its green, and when full green was under
examination the green became a yellow-green. All
through the green part of the spectrum this yellowing
was apparent, and in the blue-green, as far as Scale
No. 36 (X 5085), the colour appeared to shift in hue
towards the yellow. In the true blue-green, about
Scale No. 31 (X4886), the addition seemed to make no
difference in hue, simply making it appear rather paler.
At Scale No. 28 (X4776) the mixture of colour and
white made the blue become redder. In the violet, the
addition of white caused the colour to become redder.

These changes were interesting as throwing light on
several discrepancies which have been observed in colour
descriptions. It seemed possible that the change in hue
from near the red to the blue-green might possibly be
measured. Practically this would include by far the
most luminous portion of the spectrum.

A patch of green light at about Scale No. 40 (X 5270)
was first examined, and different percentages of white
were added to it. With the addition of 50 per cent, of
the luminosity of the white (D being 100), it was found
that an exact match in hue could be obtained by alter-
ing the colour coming through the slit of the other

R

258 RESEARCHES IN COLOUR VISION

spectrum, and that a match in luminosity could be
obtained by altering the width of the slit. The match
made indicated a hue approaching the yellow. As
less and less white was added, the match gradually
approached the scale number of the undiluted colour.

A series of colours between Scale Nos. 59*5 (X 6780)
and 36 (X 5085) were examined, with the result which
is shown in the annexed table.

It now remained to ascertain if there was any law
which could be applied to foretell the change in hue. The
first point that was evident was that the Scale No. 487
(X 5772) had something to say to any law. At this
point it is to be remembered the addition of wbite
made no alteration in the hue of the colour. On
examining Table XXXVIII., p. 239, it was at once
seen that at that point of the spectrum the proportions
of red to green were exactly those of the proportions
existing in white light.^ This seemed to give a clue to
the change in hue that takes place.

It seemed probable that the change in hue in the
region of the spectrum under investigation might be
due to the addition of the red and green sensation
luminosities contained in the white.

To. make this hypothesis and its results plain, a
reference to Table XXXVIII. should be made, in which
columns XIII., XIV., and XV. give the percentCLge
composition of the colours in terms of the red (RS.),
green (GS.), and blue (BS.) sensations.

In making tests as to the truth of the hypothesis,
the proportion of RS. to GS. in white was taken as
69 to 31, the equation for white being —

RS. GS. BS.

69-4 + 30-2 + -4.

^ As before stated, Scale No. 48*7 is the place where the red and green
curves of equal stimulation of the three sensations cut one another.

CHANGE IN HUE OF COLOUKS 259

In one set of observations the value of the full white
used was 0'6 the luminosity of D, which gave the pro-
portion of 41*4 RS. to 18 '6 GS. A rotating sector
was used to get other percentages of white.

The following is a specimen of the calculations made
for one colour with different mixtures of white : —

Scale No. 46*23 (X 5611).

RS. GS.
The proportion of RS. to GS. in luminosities in

Scale No. 46-23 is 5545 + 32-60

Addition of 60 W 41-40+18-60

Total of RS. and GS. i8 96*85 + 51-20

Converting this total into percentages, we get —

Scale No. 46*23 + 60 W - 6543 + 34*57

Turning to Table XXXVIII., columns XIII. and
XIV., we find that this would give the colour at Scale
No. 47*3. It will be seen that this is the same colour
that matched the mixture.

Taking half the white {i.e. 30 white), we get by the
same method of calculation a scale number whose per-
centage is 64'5 + 35'5. This gives a colour whose scale
number is 46*92. It will be seen that the colour matched
46-89.

Again, using one-quarter white {i.e. 15 white), we get
a percentage equation —

RS. GS.

63-1 + 36-9

which is the proportion of RS. to GS. in Scale No. 46*33
(X 5618).

The match found was 46 '3.

For all the other scale numbers mixed with different
proportions of white, the calculations were made in the

260 RESEARCHES IN COLOUR VISION

same way, and the table attached shows the results of
the matches and calculations.

Table XLV.

1 '
60 W. 90 W. 16 W.

1

7-6 W. , 8-76 W.

1*8:

w.

Scale

1

1

No.

1 1 !

1

Calc.

Found. Calc. Found. Calc. Found.

t

58-6 53-9

Calc.

Found. Calc.

Found.

Calc.

• > »

Found.

56

• • •

54-6

54*3 55-2

54*9

■ • •

54-45

« • ■

... 5219 52-6 58 5316

54

64

• ■ «

60-6

• • •

... 50-13 5012' ...

* • •

• •• • • •

• • ■

•

52-9

... 517 51-6 52-3 52-2

527

527

46-23

47-3

47-3 ,46-92 46 89 , 46*83 ; 46 3

> « ■

« • • • • •

a • •

43-65

4617

46-3 |45 45 44*6 ! 44*68 4414

44-05 ...

* ■ •

411

• • •

... '44-5 44-75 43-2 43 142

42-3 417

41*43

• • «

88-5

• • «

41-6

41*6

403

40*4 39-5

89*3

35-92

• • •

• • ••• m § m •■•

1 '

■ » •

401

40*2 '88-5

38-57

37-4

37*3

1

Another set of matches and calculated values will be
found in Table XLVI.

Table XLVI.

Scale
No.

Now.

43 W.

Calc. Found.

57*47 I 57-47
551 551

497

47

44-3

41*5

38-2

36-1

497

47

44-3

41 -5

38-2

36-1

51*76

49-34

47*58

46-2

45*36

44-8

45-2

61*82

49-34

47-6

46-16

45*36

44*83

44-8

21-5 W.

Calc. Found

52-8
49-6
47-4
45-6
44*1
42-9
43*2

10-76 W.

6*87 W.

2*68 W.

Calc.

Found.

65-96

55-06

541

• • •

54*2

• • «

44^

44*83

42-35

42-45

40-37

40*33

39-3

39-5

56-92
54-47

66-49
64-9

The results are shown graphically in the figure.

[In the above tables where there are no results given
for the greater proportions of white (as in Table XLV.
for Scale No. 56), the match became uncertain owing to
too great a proportion of white being added.]

The results show that in the parts of the spectrum
under measurement the value of the blue sensation is

CHANGE IN HUE OF COLOURS 261

unimportant as regards hue, when matches of impure
with pure colours have to be made. It is also worthy of
remark that there is a considerable range of spectrum
colours which can, by adding different proportions of
white, be caused to match in hue a pure colour. [This
points to using a spectrum in a room free from all white

Scale of Spectrum.

Fig. 86.

light, and with prisms as free from dirt as possible, to
prevent the white illumination of the surfaces.]

Another useful fact is this, that if SN. 487 (X 5772)
be matched by a mixture of red (say, at the red lithium
line) and any green, the amount of green sensation in
the green employed can be easily calculated, as the blue
has no effect on the match of hue. For although the

262 RESEARCHES IN COLOUR VISION

colour to be matched may contain blue as well as
green sensations, yet the former is accounted for in the
white (see Table XXXIX., column V.). [It must be re-
membered that SN. 4 8 '7 is very readily fixed, fi-om the
fact that it is the colour which, with pure blue, can
make a white to match the white of the crater of
the electric light, quite regardless of the yellow-spot
difficulties.]

In the determination of the equations to the three
colour sensations, one of the first researches was to find
the amount of inherent white in that coloin: which
represented the colour of the green sensation mixed only
with the white. When that was foxmd, the equation to
make white in terms of the red, green, and blue
sensations became an easy matter, and therefrom the
amount of red and green sensations was easily calculated.

In. some experiments, made with the object of checking
the amounts of red and green sensations in the colours
(see Table XXXVIII.) lying between SSN/s 48 (X 5720)
and 36 (X 5085), the place of the colour to be matched
was close to 487 and had 69-86 RS. and 30-14 GS. This
colour was isolated from one spectrum and matches
made with the colours coming through two slits placed
in the other spectrum. One of these slits was placed
at the position of the red lithium line and the other
moved about in the green as required. The matches
were made by opening or closing the slits. The
following are specimens of the results.

The red slit was placed at SSN. 59*8 (X 6501), and
the green slit at SSN. 40*8. The respective relative
luminosities through equal slits were 8*4 and 55 ; the red
slit had an aperture of 102 on an empyric scale and
the green 27.

The relative luminosities are therefore 102 x 8*4 = 85*7

CHANGE IN HUE OF COLOURS 263

and 27x55 = 148-5. Thetwo luminosities added together
= 234-2.

As SSN. 487 contains 69*86 of RS., 234-2 has to be
multiplied by this, the result being divided by 100,
which equals 163*6 of RS. (supposing, as shown before,
that BS. is negligible). But there is 857 of RS. from the
red slit. Therefore in the green colour SSN. 40*8 there
can only be 77*9 RS., the remainder of the colour being
70-6 GS.

We therefore have the colour SSN. 40*8 represented
by-

RS. GS. RS. GS.

77-9 + 70-6or 52-5 + 47-5

Turning to Table XXXVIII., columns IV. and V.,
we find that by other means SSN. 40*8 was found to .
contain 53 RS. + 47 GS.

Similarly, when the green slit was placed at SSN.
437, which has a luminosity of 73 "1 on the same scale
and a width of 27, and the red slit a width of 94, by
making similar calculations the colour (437) is repre-
sented from these observations by —

RS. GS.

577+42-3

which is the same composition as that found from the
table.

Again, with the green slit placed at SSN. 38, the
luminosity of which is 36, a match was made when this
slit had a width of 34 and the red slit of 102. The
resulting calculations gave —

RS. GS.

49 + 51

In the table it is 48*65 RS. + 5M GS. The part of
the spectrum from SSN. 64 (X 7220) to SSN. 48 (X 4720)

264 RESEARCHES IN COLOUR VISION

is readily obtained by ordinary methods, as is the portion
from SSN. 36 (X 5085) to the extreme violet. The most
difficult portion is from SSN. 48 (X 4720) to SSN. 36
(X 5085), and this can be checked by the method
indicated above.

When the white added was that of a paraffin lamp,
similar results were obtained, using the proportion in
luminosities of red to green in the white as 76 to 24.

There is no difficulty in matching one hue with
another when the two are separated by a small dark
interval. The eye instinctively ignores the blue present
in a rather remarkable manner. We shall find these
results have to be considered when considering certain
matches which are described in Chapters XXII. and
XXV.

We have one or two other questions to answer
as to the effect of the addition of white light to
a colour. One is, How much spectrum colour can be
added to white light without being perceived? Per-
haps one of the easiest methods of showing that an
appreciable quantity of coloiu* may be added to white
without being recognised is by means of a rotating disc
of white, 4 to 5 in. radius, on which equal spots ^ in.
of the colour under investigation are fastened along a
radius with intervals of, say, f in. between them, we
shall find that the outside rings which should be formed
on rotation are invisible, and that it is only the inside
spots which form a slightly coloured ring. The reverse
may also be observed if the large disc be coloured and
the spots are white. It will be found that there is a
marked difference in the results.

Using the colour patch apparatus and placing a
diaphragm to cover the outside face of the prism, and
having a slit in the focused spectrum, we have the means

CHANGE IN HUE OF COLOUKS 265

of placing a coloured spot on the square face of the cube.
The spot can then be "drowned" with white light from
the reflected white beam. (The brightness of the spot,
of course, depends on the width of the slit.) In a set of
measurements it was found that the reduced angular
apertures of the sector required to drown the colour were
as follow for the following Fraunhofer lines : —

Fraunhofer SSN. Lines. Angle of Aperture.

B 300°*

66"

D 14"

E 22°

F 160°

G 2100° ♦

The large numbers marked with an asterisk were obtained by placing
the sector in the white reflected beam. For the other numbers the sectors
were in the colours.

Taking the luminosities of the different colours
and the luminosity of the white, it was found that
between ^^ and ^ of the luminosity of the white
the colour was unrecognisable.

These results have a bearing on colour equations,
and it is only by taking a series of observations that
we get a mean equation of sufficient exactitude. The
colour equations themselves when a series is taken are
proofs of this. The mean of a large number of equations
to match white, when the red sensations were all made
equal, gave the fact that 1'5 per cent, of green could
be added without being perceived. Another series
gave 2*5 per cent., another 3*5 per cent., another 2*4
per cent. (Double the differences found were used, since
the addition of red might be made instead of green.)
The final result was that 27 per cent, of green or red
might not be perceived when the observer matched
white with the rays from the three slits.

266 RESEARCHES IN COLOUR VISION

Another plan was adopted to compare with the
above. It was to see if any change in hue could not
be observed, and to find the percentage of sensations
which the change indicated. Taking the whole spec-
trum from SSN. 56 to 29, it was found that an average
addition of 2 8 per cent, would escape notice unless
very critical examination was made. The greatest
addition that could be made without altering the hue
was found to be in the green.

CHAPTER XVIII

CONGENITAL COLOUR BLINDNESS

So far we have only considered ordinary or normal
colour vision, which is possessed by the large majority of
mankind, and it is not. a century and a half ago since any
suspicion arose that any other kind of vision existed. At
that time any departure from the normal vision was a
matter of curiosity. In the Philosophical Transactions
of the Royal Society of 1777, the case of a shoemaker
named Harris is described by a Mr. Huddart, who
travelled all the way from London to the Midlands in
order to see if all the alleged facts regarding him were
true. Harris mistook orange for green ; brown he called
black ; and he was unable to distinguish between red
fruits and the surrounding leaves. This was a case
probably of green colour blindness, as we shall see it
answers to the more exact methods now extant for
diagnosing the kind of defective colour vision.

Dalton's Colour Blindness.

At first colour blindness was called Daltonism (and
indeed is still so termed in France), from the fact that
the great chemist Dalton suffered from it, and investi-
gated the variation which existed between his and his
fellow-creatures* colour sense. It was in 1794 that
Dalton described his case. He was quite unaware of
his defect till 1792, when he was convinced of its exis-
tence from his observation of a pink geranium by candle

267

268 KESEARCHES IN COLOUR VISION

light. " The flower," he says, '* was pink ; but it appeared
to me almost an exact sky blue by day. In candle light,
however, it was astonishingly changed, not having any-
blue in it, but being what I call a red colour, which
forms a striking contrast to blue." He goes on to say
that all his friends except his brother said there was not
any striking difference in the two colours in the two
lights. He then investigated his vision by means of a
solar spectrum, and became convinced that instead of
normal colour sensations he had only two, at the most
three. These were yellow, blue, and perhaps purple. In
his yellow he included the red, orange, yellow, and green
of others, but his blue and purple coincided with theirs.
He says that " part of the image which others call red
appears to me little more than a shade or defect of light ;
after that the orange, yellow, and green seem one colour,
which descends pretty uniformly from an intense and a
rare yellow, making what I should call different shades
of yellow. The difference between the green part and
the blue part is very striking to my eye ; they seem to
be strongly contrasted. That between the blue and
purple much less so. The purple appears to be blue much
darkened and condensed" (These italics are ours.)

In what we have quoted we have a splendid descrip-
tion of a case of complete red blindness, and it has all
the advantage of having been made by a great scientific
man and observer. It is a model which may serve for
less acute observers who are similarly or less deficient in
some sensation.

Dalton further said that a florid complexion looked
blackish blue on a white ground. (He saw the blue in
the blood, and not the red.) A laurel leaf was a good
match to a stick of sealing wax. (He only saw the green
which was present in both.) Some browns he called red,

CONGENITAL COLOUR BLINDNESS 269

and others black. (The red of the spectrum was a shade
to him ; hence he called such shades red.) By the electric
light and lightning, colours appeared as in daylight ;
whilst in moonlight and candle light the colours changed
from what they appeared in daylight, but were alike.
(Moonlight is enfeebled sunlight, and the red end of the
spectrum is much enfeebled, as is also the blue and the
violet.)

Extent of Colour Blindness in the Population.

The percentage of those who do not possess the fully
developed normal colour sense is stated from statistics
to be between four to five per cent, of the male popula-
tion, and about the same number per thousand of the
female population. It is more than probable that this
is an under-statement, as the more delicate tests which
are now possible to use give a larger percentage of both
men and women who are defective.

Heredity and Colour Blindness,

The colour blindness in a healthy subject is congenital,
born with the person, and is very often hereditary. In
some cases it has been traced to exist in at least three
generations. Referring to the case of Dalton just quoted,
it is remarked that his brother agreed with him as to
the colours seen. We may presume that Dalton's father
was similarly affected. The writer has had several cases
of brothers partially colour blind, and it was invariably
found that both were deficient in the same colour sense,
but sometimes one more so than the other. In one
family, of which two members were distinguished physi-

270 RESEARCHES IN COLOUR VISION

cists, all the brothers and sisters were deficient in one
colour sense, but not to the same degree, and from what
has been stated to him the writer believes that the father
was deficient. Again, the writer knows a case in which
the father, though an artist, was colour blind, and the
son has the same kind of deficiency in his colour sense.
A case quoted further on will show that two brothers
who see no colour but light only are alike in this respect,
and presumably the defect was inherited.

As we have said before, the colour blindness of this
type is congenital. There is another class of colour
blindness which is acquired owing to disease or over-
smoking, but it carries with it in addition the loss of
form — that is, that the sight becomes indistinct. Con-
genital colour blindness is, so far as known, incurable,
whilst that caused by disease may be curable, or can
be ameliorated if treated in time.

Colour Blindness unnoticed hy the Possessor,

Colour blindness is often unnoticed by its possessor.
For instance, one gentleman of the age of seventy-four
was completely colour blind to one sensation, and yet
during all his years he had never found out that he
differed from the majority of persons in his colour sense.
His family had suspected that there was something
abnormal, owing to mistakes that he had made in recog-
nising different colom's. The writer found out what was
really wrong by his naming the red velvet seat of a
chair as black velvet. When tested in the laboratory, it
was found that one of the three colour sensations was
absolutely absent. A colour blind person may often be
told by incongruities in his or her dress. The clashing
of incongruous colours is one sign, though not quite

CONGENITAL COLOUR BLINDNESS 271

always a sure one, as it may be a love of eccentricity
which induces it.

There have been cases where a person in deep mourn-
ing has worn a bright red tie, and when taxed with the
society outrage that he had committed contended that it
was a black tie. A bright green is sometimes mistaken
for white, and the incongruities that can be committed
in such a case can be imagined. Pages might be filled
of such examples of persons who have never guessed that
their colour vision was not normal. It is sufficient to say
that the percentage of those who confess to a want of
proper colour sense is not large.

Danger of Colour Blindness.

On the railways, in the navy and mercantile marine,
colour blindness in a signalman, engine-driver, or look-
out man is a danger to the community, since the colour
of signals cannot be seen as they ought to be on a
railway ; and in the marine services neither ship's lights
nor flag signals can be correctly stated. That accidents
have happened owing to colour blindness of a railwayman
or a seaman cannot be doubted, though inquiries as
regards collisions have not brought out the facts. Owing
to the general ignorance which prevails in all grades of
society as to the mistakes that can be made by the
colour blind, it is almost unheard of that any witness
has been examined as to whether he has normal colour
vision before he gives his evidence. The often silly
remarks made by the many about colour blindness lead
one to regret that children are not taught at school that
such a defect of vision may exist, and be harmful to
the community in certain walks of life.

272 RESEARCHES IN COLOUR VISION

The Explanation of Newton s Colours in the Spectnim.

Turning to the colour sensations, we find a ready
explanation of the colours which Newton placed in the
spectrum. He saw there was a red, orange, yellow,
green (blue-green*), blue, ultramarine (he called it
indigo), and violet, and these may be taken as the general
hues seen by a normal eye.

In Table XXXIX., page 242, there are coluncms
giving the composition of the different colours from the
red to the violet. It will be seen that there is an
unbroken sensation of red from the extreme end of the
spectrum to SSN. 57, except a minute trace of green at
SSN. 58. At this number the green sensation comes into
the colour more and more to SSN. 52. The combination
of the more powerful red sensation with the green gives
a colour which may be classed as orange. From
SSN.'s 52 to 50 the green sensation is stiU more
developed, which gives a yellow. At SSN. 49 a new
factor is introduced in the shape of white, and the green
sensation becomes predominant to SSN. 38, and the
general hue is green. From SSN.'s 38 to 34 a small
quantity of blue appears with a diminishing quantity
of white, and this causes the blue-green colour. From
this number to SSN. 24, only the green and blue
sensations with white are extant, and the hue changes
to a blue. From SSN.'s 24 to 16, we have red
reappearing, and the blue sensation and white are also
present. This gives a subdivision, which may be classed
as ultramarine, whilst from SSN. 16 to the end of the
spectrum we have only the red and blue sensations in
the colour, which give rise to the violet or purple.

^ Blue-green was not in Newton's list, but it is included here, as it is a
very definite hue to those possessing normal colour Tision.

CONGENITAL COLOUR BLINDNESS 273

The following is a table of colours recognised by
normal vision when the whole spectrum is viewed. A
large number of persons were examined, and the mean
beginning and end of the eight colours are given.

Table XLVII.

From Naked Spectrum.

From

Diagram.

Beginning.

End.

Beginning.

End.

Red .

) End of
I spectrum

I to 55

1 End of )
) spectrum J

to

57

Orange .

55

„ 51

57

>y

60-5

YeUow .

51

49

50-5

i>

48-5

Green .

49

37

48-5

ft

37-5

Blue-green .

37

„ 34

37-5

>»

34-5

Blue .

34

» 24

34-5

t9

24

Ultramarine .

24

18

24

tt

16

Violet .

18

5 end of )
" ( spectrum. ]

16

tf

I

end of
spectrum.

The boundaries of the colours viewed in the naked
spectrum are undefined, one colour blending into
another ; that the similarity of the diagrammatic and
observed boundaries are so nearly alike, is somewhat
remarkable.

Normal Spectinim Colours as seen by the

Colour Blmd.

It is interesting to show the colours which to the
normal eye represent the white of the colour blind.
Let three slits be placed in the spectrum : one at the
position of the red lithium line, another at SSN. 37 '5
(for the ordinary arc light), and the third at the
position of the blue lithium line. Let the normal eye
match the white of the arc light with the mixture
of the rays coming through the three slits. If now we
cover up the red slit, the colour on the screen will be a

sea green, and this will match the white of the red blind.

s

274 RESEARCHES IN COLOUR VISION

Similarly, covering up the green slit we get a purple
which matches the white of the completely green blind.
(We can cover up the blue slit and we shall have the
white of a blue blind, but as such blindness is almost
unknown, it is not necessary to deal further with this
form of blindness.) If a person is half red blind, we
shall get the colour of his white by closing the red slit
to half its aperture, and so with the green blind. For
other factors of colour blindness we have to close the
slits, multiplying the aperture by the factor. If the
colour blindness is incomplete, we may expect that in
naming, the colours to the normal eye may differ con-
siderably from those which they appear to the latter.
Thus a partially red blind would be apt to class the
yellow as greenish and the scarlet as yellow ; the ex-
treme end of the red would not be perceived at all,
unless the spectrum were very bright ; the limits of the
green would also vary, and, in bad cases of red deficiency,
the violet would become a blue. In cases of partial
green blindness, the yellow might be called orange or
even red, and the blue-green would be classed as green.
In very bad cases of green deficiency, the whole of the
spectrum from the yellow to the blue might be called
white or grey, as the amount of their white in the inter-
mediate regions would shroud the colours which they
could see if deprived of the white. In naming the
colours of the spectrum, it must be remembered that
names are learnt from the normal eye's perceptions, and
it is the endeavour of the colour blind to call the
different parts of the spectrum by the appropriate
names from recollection of colours which they see in
everyday life, and which are named by normal vision.
The colour blind's judgment is often formed by the
luminosity of a colour, and not by any marked difference

PLATE 1.

•025 G8.

Completely
green blind

Completely
red blind

Spectrum colours as named by persons who were completely or
nearly completely red or green blind.

Normal

PLATE //.
30 40

Normal

J L

J L

Dots of
pure colour

Patches of
pure colour

T r

spectrum colours as named by a person who possessed

•05 of Red Sensation.

Tlic whole
spectrum

CONGENITAL COLOUR BLINDNESS 275

in the hue, as is the case with the normal eye This
being the case, we may expect (and our expectations
are usually realised) that under varying circumstances
the colour blind will give various names to the same
(normal) colour.

Plate I. illustrates the colours which persons com-
pletely and one nearly completely colour blind name the
spectrum colours.

There has been no endeavour in these diagrams to
give any idea of the luminosities of the different colours,
but only the hues which the colour blind say they see.

Plate I.

In Plate I. the bottom figure shows graphically how the

normal eye sees the spectrum.
No. 2 line shows how the completely red blind sees the

spectrum.
No. 3 line shows how the completely green blind sees the

spectrum.
No. 4 is a case of nearly complete green blindness.

Plate II.

No. 1 line is the normal spectrum shown graphically.

No. 4 shows the naked spectrum colours as described by
a person who was largely deficient in red sensa-
tion.

In 2 line are the names which he gave to dots of pure
spectrum colours about ^q of an inch in diameter
when standing about 16 feet away from them.

In 3 line are the names which he gave to individual
patches when pure spectrum colours were shown
to him.

CHAPTER XIX

COMPLETE RED AND GREEN COLOUR BLINDNESS*

It will perhaps be easier for the reader if we describe
what has been found to be the deficiencies in perception.
Turning to Fig. 98 of the last chapter, we have the three
sensations for the normal eye shown in terms of equal
stimulation for the three perceiving apparatus.

The Normal Sensations which are absent to the

Colour Blind.

If one of these sensations is absent, say the red, in
the first instance, what eflTect should it have on the
recognition of the different colours of the spectrum ?

In the first place, from SSN.'s 60 to 65 there will be no
sensation of colour, as in that region only the red should
be stimulated, and there is no red apparatus to stimulate.
Between SSN.'s 50 to 60 there will only be the green
sensation, and that will be felt in a purity that the
normal unfatigued eye cannot feel. All the colours from
the scarlet to the yellow, to the red blind, will be
different intensities of the green sensation.

At SSN. 49 the blue sensation will begin to be felt.
Taking a forward step, let us see what the sensation of
added blue means to the red blind. At SSN. 34*6 the
green and blue curves cut one another ; and as the
ordinates at the point of intersection are equal, the colour

' See Papers Nos. 5 and 6.

276

COMPLETE RED AND GREEN BLINDNESS 277

which to the normal eye is green will appear to be a white
similar in hue to that which forms the spectrum, and can
be matched with it by the red blind. The addition of
blue from SSN/s 49 to 34*6 means that the green sensa-
tion begins to be slightly paler at 49, and the paleness
increases until at 34*6 all the colour has gone. From
SS N. 3 4*6 to SSN. 16 the green sensation diminishes
grsl^l^l^ whilst the blue increases, so that at, say,
SN.^3, there is white, to which a little blue has been
acUled, and the blue increases in purity until SSN. 16 is
reached, when there is no admixture of green at all.

Theoretically, then, the absence of the red sensation
means that there are only two sensations which in the
centre of the spectrum are more or less contaminated
with white. If a red blind be asked to name the colours
of the spectrum, he wiU name them as stated above,
though he may call the green yellow ; but this is rarely
the case, and has no significance, being merely a ques-
tion of nomenclature. To the totally red blind person
the spectrum is shortened at the red end, and he sees
only green, and blue diluted with his white, the white
being a mixture in definite proportions of green and blue.
If, then, we find anyone who cannot see the red from
SSN.'s 60 to 65, we shall diagnose that he is red blind. It
must always be difiicult for a person with normal vision
to interpret the descriptions which colour blind people
give of the spectrum.^ The majority of the persons they
associate with have normal vision, and they educate
themselves to recognise and name the colours as named
by this majority, judging not by the hue, but by the
shades and purity of the two sensations they possess.
It is this system of self-education that breaks down when

' It is less difficult for persons who carry out experiments in colour
fatigue of the retina (see Chapter XXV.).

278 RESEARCHES IN COLOUR VISION

proper tests are applied. The kind of tests which lead
to the certainty of the detection of colour blindness will
be given later.

When there is complete green blindness, we can
ascertain what theoretically would happen when such a
colour blind describes the spectrum. In the first place,
the spectrum would be of the same length as it is to normal
vision. Between SSN.'s 50 and 65 red only would be felt,
but in different shades, the maximum brightness being
at about SSN. 52. At 49 blue should begin to be felt ;
and will gradually increase as the normal full green is
approached. At 37 '5 in the green the two equal area
curves cut one another, so at this point of the spectrum he
should see a white which would match that of which the
spectrum is formed. From this point to SSN. 49 the
spectrum colours should be to him red mixed in gradually
diminishing quantities with the (green blind) white. On
the more refrangible side of 37*5, the (green blind) white
would be mixed with violet in gradually increasing
quantities till SSN. 14, where the relative amounts of
the red and blue sensations remain the same. When
a green blind is asked to name the various colours of
the spectrum, he may call the red sensation yellow,
red, or green, and he may from education even name the
various colours correctly, but tests with the spectrum
will soon convince the examiner that what he theoreti-
cally ought to see he does see, and that the foregoing
description is correct.

Reverting to what the green blind calls white, it was
shown in the last chapter that his white is a brilliant
purple, and yet we have just stated that there is a point
in the greeii of the spectrum which to him is a match to
the white.

A glance at the diagram, p. 240, will explain this

COMPLETE RED AND GREEN BLINDNESS 279

apparent anomaly. At SSN. 37 '5 there is no green
sensation as felt by the normal eye, and the only sen-
sations felt by the completely green blind are blue and
red, which when mixed give the purple of the experi-
ment described in the last chapter.

Luminosity of the Spectrum to the Colour Blind.

It is quite as easy, indeed it is easier, for the com-
pletely colour blind to measure the luminosity of the
spectrum than it is for the normal eye, as there are only
two sensations instead of three to deal with, and there
is one place for each kind where the spectrum matches
exactly their white.

In the trichromatic theory of colour vision, the three
sensations of red, green, and blue are each totally dis-
tinct, and in complete green or red blindness one of
these two sensations is totally absent. It therefore
follows, if this theory is not merely a working hypothesis,
the luminosity curve of the red blind, if added to that
of the green blind, when the maximum numbers given
in cols. XIIL and XIV., Table XXXVIII. (p. 239), are
taken as maxima, should give the luminosity curve of
normal colour vision, with one luminosity curve of the blue
sensation in addition. For red blind luminosity is com-
posed of green sensation + blue sensation, the green
blind luminosity of red sensation + blue sensation, and
the normal colour vision curve of all three sensations.
By the addition of the red and green blind luminosity
curves, we should have that of normal colour vision curve,
together with an extra blue sensation. The luminosity
of the blue sensation is very small compared with those
of the other two, and may vary slightly, as said before,
owing to difference in the absorption by the yellow spot,

280 RESEARCHES IN COLOUR VISION

80 that roughly the addition of the red blind and green
blind curves should be very close to the curve of
nonnal vision.

Table XLVIII. — Luminosities of three Completely Chreen Blind and
four Completely Red Blind, The Mean Luminosities of the Red
and Green Blind are abided together and compared with the
Luminosity of the Normal Colour Vision^ to icTiieh an extra Blue
Sensation Luminosity is culded.

1

Lamiaoaity

1

Addition

of NorUMl

Oreen blind.

RedbUnd.

of IV.

Colour

and IX.

Vision
+BS.

bSN.

X.
6957

I.

K.
2

11.
F.

IIL

D.

2

IV.

V.
G.

• ■ ■

VI.
H.

• • •

VII. VIII.
K. L.

• •• ■ « •

IX.

X.

XI.

Mean.

Mean.

1

62

2

2

• • ■

2

2

60

6728

7

7-6

7-2

7-3

• • •

• • •

• • • « • •

• ••

7-3

• 1

68

6521

20

21-8

22-2

21-3

• ■ •

• ••

• • • ■ ■ •

• ■ •

21-3

21

56

6330

46

48

46

467

3-3

3

2-8 3

3

497

50

64

6152

75

71

72-2

727

81

7-5

7-1 ; 7

7-4

801

80

52

5996

80-3

81-6

80*6

80-5

16-6

16-5

13-7 17

15-9

96-4

96

50

5850

75

77

74-4

75-5

27

24

26-2 25

25-5

101

100 1

48

6720

67

64*8

661

66

31-4

30

32 30

30-8

96-8

97

46

5596

56

55«

55-5

557

32-8 32-5

33 33

32-8

88-5

87

44

5481

45

46

45-5

45-6

30-5 32

31*2 I 32

317

77^

75

42

5373

35-5

35

35-5

358

27 i28

28*3 28

27-8

631

62-5

40

5270

271

26-4

26*5

267

21-4 1 24-6

23-8 23

1

23

497

50

38

5172

16*5

18

18

17-6

15-2

17

181

17

16-8

84-3

36:1

36

5085

10

11

11-9

10-9

9-5 11-6

12-5

10

10-9

21-8

241 ,

34

5002

6-2

6*2

7-2

6-5

6

6-5

6*8

6-8

6-6

13

14-3 1

32

4924

4-5

S'4

5

4-3

3-8

4

4*4

4-4

3-9

8-2

8-6

ao

4848

3-4

2-5

3-6

3-2

2-4

2-8

2-5

3-2

27

5-9

6-9

28

4776

8

2

2-8

2-6

22 2

2

2-6

2-2

4-8

4-2

26

4707

2-5

1-65

21

21

1-9

1

1-2

1-9

1-5

3-5

3

24

4680

2

1-30

1-5

1-6

1-4

7

•8

1

1

2-6

2-2

22

4578

17

1

1-2

13

11

•5

•5

•7

7

2

1-65

20

4517

1-5

75

1

1

•8

•3

•3 -3

■4

1-5

1-33

A large number of curves ^ have been plotted by the
writer from observations of luminosity made by both

* See Paper No. 21.

COMPLETE RED AND GREEN BLINDNESS 281

kinds of complete colour blindness. Luminosity curves
are showu in Fig. 87, and the table gives the measures
made. In the case of the red blind the maximum of
brightness is at about S8N. 46, and the curve of lumi-
nosity is made at that point to have an ordinate of 32'8.
Similarly, the green blind has a maximum at SSN. 52,

and at this point the ordinate is made to have a height
of 80'6, whilst the normal vision curve has a maximum
near SSN. 50, where it has an ordinate of 100.

These numbers for the masima of the red and green
blind are those found for the red and green sensation in
Table XXXVIII. That it is justiBable to use these

282 RESEARCHES IN COLOUR VISION

numbers will be shown by the red and green blind ** ex-
tinction of light" curves (see p. 291).

Comparing together columns X. and XI., it will be
seen that they agree together, And that any small diflTer-
ence is accounted for by errors of observation of eight
persons in all, seven of whom were unacquainted with
the method of measuring the luminosity of colour until
their luminosity curves were taken.

Details of the Measurements of the Green Sensatioyi

Curve by a Red Blind,

We give in some detail the finding of the luminosities
of the green sensation existing in the different colours by
an observer who was totally red blind. ^'

These observations were the matching in luminosity
and " hue " of a patch of white light by a mixture of
two colours, one on each side of the "neutral" point.
Two standard places in the spectrum were chosen, in
each of which was placed a slit — one in the red, in
which it was known that the blue sensation was absent,
though the green sensation was present, and the other
in the violet, in that position in which the green sensa-
tion was absent.

The relative luminosities of these two rays when
passing through equal apertures of slits was determined
by X. : that in the red (SSN. 56-82) being 2, and that in
the violet (SSN. 911) 014. These luminosities, though
taken on a different day to those on which the luminosity
curves were taken, agree well with the luminosities
shown by the curve at these points.

The observations were made as follows. The slits
were first of all kept in the standard places, and a series

^ See Paper No. 26.

COMPLETE KED AND GREEN BLINDNESS 283

of matches made with the white by opening or closing
the slits till the right hue was acquired. The luminosity
of the white patch, when it matched in luminosity the
mixed colours (the two patches being in contact with
one another, each being f inch square), was measm'ed
by introducing into the path of the beam forming it
sectors the apertures of which opened and closed at
pleasure during rotation. The aperture of the sector
indicated the white luminosity. The relative widths of
the slits were measured by placing a lens of very short
focus in the path of one of the slits. This gave a magni-
fied image of the aperture on a distant screen on which
a ^-mm. scale was fastened. When the aperture of one
slit was measured, the slide in the spectrum carrying the
slits was moved, so that the second slit was illuminated
by the same colour and its aperture measured. The
slide was then moved back to the position it first
occupied, the small lens moved away, and fresh
readings were taken. (Care was taken that the small
lens always occupied the same place in relation to
the first slit when it had to be replaced.) When a
series of observations with the slits in the standard
positions had been made, the red slit was moved to the
sodium D light and a fresh series made with the first
slit in that position and the second in the standard
position in the violet. A series of readings was made
as before. The red slit was then moved into various
positions between SSN. 56*8 and the neutral point, the
violet slit remained fixed, and matches were made with
the white. When this was finished, the red slit was
placed at D and matches of white made with the violet
slit, when in different parts of the spectrum, on the more
refrangible side of the spectrum. (The D light was
chosen for the red slit, as it contained a larger

284 RESEARCHES IN COLOUR VISION

amount of green sensation than the standard position,
which was convenient.) Where the width of either or
both of the slits was very small, the aperture to be
measured was increased by placing in the path of one or
both of the rays a small cardboard sector with fixed
apertures. After measuring the apertures, they were
one or both diminished according to the aperture of the
cardboard sector.

The method by which the composition of the different
rays was determined is shown below, two examples
illustrating it.

The red slit was placed at SSN. 48 8, the violet
slit being at the standard place SSN. 9*11.

The equation to match white was, in terms of slit
apertures —

(48-8). (91 1). White.

(i.) 41 + 106 = 55

Increasing this equation to make 100 white, we
have —

(48-8). (911). White

(ii.) 75 + 193 = 100

The standard equation with SSN.'s 5682 and 9*11,
in terms of slit apertures, had been found to be —

(56-82). (9-11).

(iii.) 1116 + 228 = 100
Equating (ii.) and (iii.), we get —

(48-8) (56-82). (9-11).

75 = 1116 + 35

Multiplying the right-hand members by 2 and 0*14
respectively, we get, after dividing by 75, the luminosity
of SSN. 48-8 as—

GS. BS.

29-9 + 0065

COMPLETE KED AND GREEN BLINDNESS 285

in luminosities (GS. and BS. being used as the symbols
of green and blue sensations).

Again, for SSN. 46*23 we have the following equa-
tion : —

(46-23). (9-11). W.

18 + 46 = 25
or

(46-23). (9-11). W.

72 + 184 = 100

Equating this with (iii.) and converting the slit
apertures into luminosities, we get —

GS. BS.
Luminosity of SSN. 46-23 = 315 + 0*085

In this manner the luminosities of the different
wave-lengths to X. were worked out.

The following is a table of the final determina-
tions : —

Table XLIX.

SSN.

GS. BS.

SSN.

*

GS. BS.

54-27

- 7-6 +0-003

36-62 =

11-2 +0-093

60-6

= 22 +0005

30-22 =

4-06+0165

48-8^

=29-9 +0-066

2601 =

0-77 + 0-238

46-23 »

=31 -5 +0086

19-71 =

011+0-262

40-92 »

=26-5 +0086

14-39 =

+0-203

38-62

= 19-2+0-068

911 =

+014

These figures were plotted and a curve drawn through
the points. The following table was then constructed
from the curves.

1 The blue sensation curve is like that of the normal curve as far as
38 '62 ; below that it differs, but the amount of blue is so small in the equa-
tion that it may be possibly different when repeated observations are made.

286 RESEARCHES IN COLOUR VISION

Table L. — TaJble showing X.'s Sensation Curves as Luminosities ; also
the same CwTes from Phil. Trans. ; also X!s Total Luminosity Curve
taken direct.

I.

IL

III.

IV.

V.

VI.

VII.

VIII.

1 IX.

XL's lensation

Coloor sensation In

cnnres in

luminosities from

X.'s

%

laminosities

X.g

Table XXXVIII.

Normal

luminosity

88N.

X.

08. +B8.
added.

1

GS. + R.S
•added.

curve
taken
direct.

GS.

BS.

OS.

B8.

58

6531

1

1

1 • ••

•21

• • ■

•21

•2

56

6330

2-5

■ •■

2-6

2-25

• • «

2-25

2-25

54

6152

7-2

trace

72

7-60

■ • •

7-60

75

52

5996

15

tmce

15

15-36

■ • •

15*36

151

50

5850

25

•024

26 02

25

• • •

25

26

48

5720

31

•062

31-06

31-78

-020

3180

32

46

5596

32-5

•087

32-59

32-70

•027

32-73

32*5

44

5481

31-5

•100

31-6

3130

•032

3133

31-5

42

5873

29-2

•093

29-3

2775

■042

27^80

27*5

40

5270

25

•060

25-8

24-09

•058

2415

24

38

5172

18-5

■070

18-57

18-43

•083

18 52

18^5

36

5085

12

•090

12-09

12^83

•101

12-90

13

34

5002

8-3

•110

841

7-80

•124

7-98

7-5

32

4924

5-5

•134

6-63

4-77

•145

4-92

4-5

30

4848

3-5

•160

3-66

3-08

•174

3-83

3

28

4776

2-2

•190

239

203 ,

-202

2^23

2

26

4707

1-2

•220

1-42

116

-243

1^39

1-2

24

4639

•5

•250

•75

•53

•262

79

•96

22

4578

•3

•255

•55

•27

•247

•52

75

20

4517

•11

•253

•36

•10

-234

•33

•65

18

4459

•04

•242

•28

-04

•202

•24

•42

16

4404

• ••

•224

•22

•01

•180

•19

•25

14

4349

• • •

•195

•19

■ • «

-154

•15

•22

12

4296

• ••

•176

•17

« « •

•126

-13

•2

10

4245

• • ■

•150

•16

a • •

•098

•10

•17

8

4198

• • •

•130

•13

■ • •

•073

•07

•126

Column I. is the Standard Scale No. (SSN.), column
II. is X, columns III. and IV. the green and blue sensa-
tion curves derived from X.'s equations, column V. his
luminosity curve by the addition of III. and IV.,
columns VI. and VII. are the curves of the green and
blue sensations taken from Table XXXVIII., column
VIII. is the luminosity derived from the addition of

COMPLETE RED AND GREEN BLINDNESS 287

VL and VII., column IX. is X.'s luminosity curve taken
direct and reduced as before described.

The results obtained from the measures made by X.
are valuable. It has frequently been asserted that
when luminosities are measured in the manner described
in Chapter VIII., something is measured which is not
luminosity. Now X., when he made his colour equations,
matched the white with the rays coming through dif-

Fio. 88.— X.'a colour curves. (Red bliod.)

ferent apei-tures of slits, and the only luminosity he
measured was the luminosity of the two white patches,
to which no objection can be raised. It was only when
these readings had been made that the question of
luminosity of his colours entered into the problem.
Only two luminosities of coloured rays were measured,
and these were applied to his slit apertures to find
the luminosity of the different rays. As mentioned
before, the luminosity measured direct and that derived
from the equations are practically identical, so that a
totally different kind of measurement confirms the direct
method of measuring the luminosity.

We win now take D.'s luminosity curve and X.'s

288 RESEARCHES IN COLOUR VISION

green luminositj curve only, which should give, when
added together, the normal colour viaon curve cloeelj,
as only one blue sensation curve will be found in the
compounded curve. Table 11. gives the results. The
comparison of the compounded curve with that taken
direct by the normal colour vision eye shows how closely
they are alike, and the similarity is very remarkable,
considering that the observations of three different per-
sons are used.

Tablk LT.

I.

II.

in.

IV.

V.
HLaiid TV.

2

VI. ;

88N.

X.

lominority.

WBMfciOD

lamtBoiity.

1

Nomud
lnniinoaiij.

1

62

1

6d57

2

« • •

2 '

60

6728

7-2

■ « •

7-2

7

58

6521

22-2

1

23-2

21

56

' 6330

46

2-5

48-6

60

1 54 1

6152

72-2

7-2

79-4

80

52

5996

80-6 1

15

96-1

96

50

5850 '

74-4

25

99-4

100

48

5720 ,

661

31

971

97

46

5596

55-5

32-5

88

87

44

5481

45-5

31-5

77

75

42

5373

35-5

292

64-7

62-5

40

5270

26-5

25

51-5

50

38

5172

18

18-6

36-5

36

36

5085

11-9

12

239

24

34

5002

7-2

83

155

14-2

32

4924

5

5-5

10-5

8-5

30

4848

3-6

35

7-1

5-7

28

4776

2-8

2-2

5

4

26

4707

21

1-2

33

28

24

4630

1-5

•5

2

1-9

22

4578

1-2

•3

1-5

1-4

20

4517

103

•11

114

1-1

18

4459

•72

•04

•76

•86

16

4404

•62

« • «

•62

•7

14

4349

•52

• • •

•52

•56

12

4296

•42

• • •

■43

•45

10

4245

•34

• ■ «

•34

•34

COMPLETE BED AND GREEN BLINDNESS 289

In the table, column I. is the SSN., column II. the
wave-length, column III. shows D.'s luminosity curve,
column IV. is the green sensation of X. in luminosity,
column V. gives the results of the addition of D/s
luminosity to X.'s, whilst column VI. shows the
luminosity curve for normal colour vision. Columns V.
and VI. have to be compared together to test the
strength of the theory.

Extinction of Light by the Completely Red and

Green Blind.

The extinction of light from the spectrum colours ^
by completely red and green blind eyes to obtain
measures of the total quantity of light which they see
compared with an eye having normal vision now becomes
necessary. By such measures we ought to confirm the
maximum luminosity of the spectrum to the completely
red and green blind relatively to the normal as shown
in the sensation curves at p. 239. The table gives
a specimen of the extinction of light in millionths of
the limiinosity when the D light has a luminosity of
1 candle at 1 foot to normal vision. In both examples
the maximum ordinates in the luminosity curves of the
green and red blind have been made 80'6 and 33
respectively. If under these conditions the extinction
of the blue end of the spectrum is the same for the
green blind and normal, since they both have sensations
of red and blue in that region, and if the red blind
shows a proper ratio for his extinction values in the
same region compared with the normal, we have the
strongest evidence that these values for the maxima of
the two kinds of the complete colour blinds are correct.

^ See Papers Nos. 4 and 21.

T

290 RESEAKCHES IN COLOUR VISION

Table LTI. — Extinction of different Cotovrt of the Speetrum liy a Green
blind and Red blind.

I"-

^

e>^T«>t

1^

VZi"

S-

Villon] In

■DllUonUu

a nan
blind.

"•SS?*"

1

nortty.

6728

1200

7-2

6521

650

22-2

6330

260

46

6162

IM

72-2

6996

66

80-6

5850

S5

74'4

5720

12-5

661

6696

7-5

&5-5

5481

5-6

45-5

5373

5

35-5

5270

5

^5

6172

6

1

6085

6

11-9

5002

7

7-23

4924

9

5D5

4848

I2'5

3-61

4776

17

2-79

4707

25

2-07

4630

34

1-55

4578

45

1-24

4r>i7

75

103

4459

126

■72

4404

S06

■62

4349

225

■52

4296

270

■43

424I>

320

■34

mllliantbi ■'TS',''' *

iH Table j

Columns V. and VIII. are the most important.
They are obtained by multiplying the luminosity by

COMPLETE RED AND GREEN BLINDNESS 291

the extinction and dividing by 100, which gives the
extinction value when every ray is made of the lumi-
nosity of 1 candle at 1 foot. The figure gives this value
for the red and green blind and for normal vision. It
shows the difference in the extinction values.

Fla. 89. — Bed and Green Blind ExtJDOtlon Currea, each ray bavlug originally to
tbem the ImniDOsit; of one candle. Normal vlaion eitinctioD :■ ibown ae
a dotted line.

The extinctions for the "one-candle luminosity" of
each ray is practically the same in the violet for the
normal colour vision as for the green blind. There is no
reason why it should be different, since there is the

292 RESEARCHES IN COLOUR VISION

same proportion of red to blue sensation in both of
them. This is only arrived at by making the normal
colour vision and the green blind maxima of luminosity
100 and 806, which is that adopted from the sensation
curves. The red blind shows an extinction more than
three and a half times that of the green blind. In this
case it has to be remembered that in this part of the
spectrum the value of the red sensation is to that of
the blue as 100 to 28. As the red blind have no red

sensation, the extinction value should be or 3" 57

greater. The agreement is fairly complete, but this
again requires that the maximum luminosity of the red
blind should be 32*8 when that of normal colour vision
is 100, the same as that derived from the colour sensa-
tion curves. We are thus led to the conclusion that
when the same white light falls on the retinae of the
colour blind, they suffer in the luminosity stimulated
compared with normal vision. The relative areas of the
luminosity curves are nearly —

830 for normal vision ;
580 for green blind ; and
250 for red blind.

If the normal vision has an impression of . .100
the green blind has but . . . . .70
and the red blind only close upon . .30

This looks as if the colour blind were at a disadvantage
in regard to the appreciation of light as a whole.

CHAPTER XX

INCOMPLETE RED AND GREEN COLOUR BLINDNESS

Besides the cases of complete colour blindness which we
considered in the last chapter, there are still more
numerous cases of what are called by some abnormal
trichromatic vision,^ but which it is preferable to call
incomplete colour blindness, in which one of the pieces
of apparatus in the eye is only partially sensitive.

Similarity of Sensation Curves in the Red and
Green Blind compared ivith the Normal.

As far as incomplete blindness has come under the
writer s observation, the luminosity curves of the red and
green sensations are similar (in a mathematical sense) to
those existing in normal vision — that is to say, if in the
normal (say) red curve an ordinate of one colour indicates
a perception of ** a" red, and for the incomplete red blind
a perception of "6" red, then in any other position in
the spectrum that is not affected by yellow spot differ-
ences in absorption (so long as the luminosity does not
come under the category of that of a feeble spectrum), the
proportion of normal red perception to those of incomplete
red blindness is as a : 6. A reference to Table XXXVIII.
will show why the place of maximum luminosity travels
from SSN. 50 and SSN. 46 as blindness becomes more and
more pronounced. The following table (Table LIII. ) and

* See Paper No. 22.

293

294 RESEARCHES IN COLOUR VISION

diagram give the luminosity curves for eyes which only
perceive one-third of the red sensation and one-third of
the green sensation. In the first the maximum is closely
at SSN. 48 (X 5720), and in the second at SSN. 51

Table LI 1 1. — Sliowing the Calculated Luminosities of Incomplete Red

and Green Blindness,

Normal

Bed blind.

Green blind.

88N.

X.

lamt-

Loml-

Lamt-

notitj.

DOSttj

of red

}BA.

as.

BS.

notity
of green

R8.

ioa.

BS.

64

1

blind.

blind.

7217

•5

•2

•2

• ••

• « ■

•6

•5

■ » ■

« ■ «

62

6957

2

•66

•66

■ ■ •

■ * •

2

2

• ■ *

• ■•

60

6728

7

2-83

2-33

...

• * ■

7

7

• • •

• • ■

58

6521

21

7-14

6^93

•21

• • •

20-9

208

7

• » •

56

6330

50

181

15^9

2*25

• • •

48-6

47-8

•75

■ « •

54

6152

80

31-7

241

76

• • «

74-9

724

2-5

• ••

52

5996

96

42-3

26^9

154

• ■ ■

85 •S

80*6

5-2

w • •

50

5850

100

60

25

25

• ■ •

83 3

75

83

• * »

48

5720

97

53-5

217

31-8

•03

75^9

65-2

106

•03

46

5596

87

50-9

18-1

327

•1

65

54-2

107

•1

44

5481

75

46-9

14-5

313

•12

541

43^6

104

•12

42

5373

62*5

39 3

11-5

277

•12

40^9

31-6

9-2

•12

40

5270

60

82-8

8-6

24 1

•11

33^9

258

8

•11

38

5172

36

24-3

5-8

16-4

•09

237

17-5

61

•09

86

5085

24

16-6

37

12^8

•1

15-5

11 -1

4^3

•1

34

5002

14-2

10

21

7-8

•12

8-9

6-2

2-6

•12

82

4924

8-5

61

1-2

4^8

•14

5-3

3*6

1-6

•14

30

4848

57

4-07

•82

3-08

•17

3-65

245

l^OS

•17

28

4776

4

272

•59

203

•2

263

176

•67

•2

26

4707

2-8

186

•47

115

•24

2 03

141

•38

•24

24

4639

2

117

•38

•53

•26

159

116

•18

-26

22

4578

1-4

•82

•3

•27

•25

126

•81

•09

•25

20

4517

11

•69

•26

•1

•23

1^08

•77

•08

•23

18

4459

•86

•45

•21 .

•04

■2

•83

•62

•01

•2

16

4404

•7

•36

•17

•01

•18

•69

•51

• « •

•18

14

4349

•56

•284

•131

■ » •

•154

•546

•392

• • •

•154

12

4296

•46

•237

•111

• ••

•126

•46

•334

* • •

'126

10

4245

•35

•182

•084

■ ■ •

•098

•351

•263

• •»

■098

(X 5922). The maximum at SSN. 49 (\ 5873) is reached
when the red sensation is about two-thirds of the normal,
at SSN. (47) (X 5658) when it is about one-tenth of the
normal, and at SSN. 46 (X 5600) when there is no red

INCOMPLETE RED AND GREEN BLINDNESS 295

seosation. In the green blincl, when there is no green
sensation, the maximum is closely at SSN. 52 (X 6000).

Au67-2i B, 61-S; Red lithium. 69-8: C, 681; D, 50-6: E,39'8:
Blue lithium, 22-8; G. 11-2; H,-7'l.

Thus by observing the position of maximum lumi-
nosity we can form an approximate diagnosis of the
amount of the defect and as to the sensation in which
the defect exists.

First Method of Ascertaining the Amount of
Colour Blindness-
Suppose that we have a luminosity curve taken by
(say) an incompletely red blind eye, the question comes
whether we can find still more exactly than by the posi-
tion of the maximum ordinate the amount of deficiency
that exists.

If by any means we can make the ordinates of the
luminosity of any ray obtained by the colour blind of

296 RESEARCHES IN COLOUR VISION

proper height when such ordinate is compared with that
obtained by normal vision, we can then compare all the
ordinates of the curve given by the former with those given
by the latter, for both curves will be on the same scale. If
the trichromatic theory holds good, then the difference
between the ordinates of the normal and the colour blind
curves (say of an incompletely red blind) should, at every
place in the spectrum (except may be in the blue), give a
curve which is mathematically similar to the normal red
sensation curve. The ratio of the ordinates of this curve
to the ordinates of the normal red sensation curve will
give the amount of red sensation deficient in the incom-
pletely red blind eye.

When the incomplete blindness is to the green
sensation, the same line of argument applies.

For convenience of reference, the table on p. 297 has
been extracted from Table XXXVIII.

Two cases, one of incomplete red and the other of
incomplete green blindness, will now be given. The
luminosity measiu'es were taken several years ago, and
before the three sensation curves of the writer's (normal)
eye had been found. (Without knowing whether a com-
parison of the luminosities to the colour blind eye of the
spectral colours with those of a normal eye when using
the same white would be of any value, in some cases
measures by both were made and recorded. To these
we shall refer later.)

It must be again pointed out that, owing to differences
in the absorption by the macula lutea in different eyes,
the blue sensation curve may not always be capable of
the same treatment as the green or red sensation curves.
But from the red end of the spectrum to about SSN. 40
(X 5270) ^ this variation will not appreciably affect the
results.

^ See Paper No. 4.

INCOMPLETE RED AND GREEN BLINDNESS 297

Table LIV. — Shomng the Composition of the different Rays of the Spec-
truniy the Spectrum being foitned of the light from the Arc with
Sloping Carbons,

L

11.

III.

Luini*
nosity of
spectrum.

IV.

V.

VI.

VII.

VIII.

IX.

Standard

Scale

No.

(SSN.).

X.

Percentafire compotiUon

of colours in terms

of sensation.

Luminosity of sensation.

BS.

OS.

• • •

BS.

1

BS.

OS.

• « •

BS.

• ■ •

64

7217

•6

100

1

• • •

•5

62

6957

2

100

• K •

• • •

2

■ • ■

• • •

60

6728

7

100

• • •

• ■ •

7

m ••

...

58

6521

21

99

1

■ • •

2079

•21

• • •

66

6330

50

95 5

4-5

• • •

4775

2-26

• • •

54

6152

80

90-5

9-5

• • ■

72-4

7-6

• • •

52

5996

96

84*2

15^8

• • a

80-64

1536

• * •

50

6850

100

76

25

• • «

75

25

• • •

48

6720

97

671

33

•02

6616

32-01

-019

46

5596

87

62

37 9

•081

5406

32-97

•027

44

6481

75

57-2

419

•042

48-8

31-5

•032

42

6373

62*5

55

44-9

•167

34 4

28-06

•042

40

6270

50

51-8

48-6

•117

25-61

24-3

•058

38

5172

36

48-5

513

•23

16-51

18-4

•083

86

5085

24

46*08

53-5

•42

1109

12-83

•101

34

5002

14-2

4379

66-34

•87

6-22

7^86

•124

32

4924

8-5

4217

56-13

17

8-68

477

•145

30

4848

6-7

42*24

54 6

3-16

2-45

806

•174

38

4776

4

44-36

60-54

5-2

1-76

203

•202

26

4707

2-8

50-02

41-3

8-68

1-41

115

•243

24

4639

1-95

6866

28

13-44

115

■63

•262

22

4578

1-4

65-56

16^3

17-64

•91

•27

•247

20

4617

11

7072

8

21-28

77

•1

•234

18

4469

•86

71 •88

4-6

23-62

•62

•04

•202

16

4404

7

72

2

25-76

-51

•01

•18

14

4349

•56

72

•6

2744

•392

• • •

•154

12

4296

•45

72

• • •

28

•334

■ • •

•126

10

4246

-35

72

• ••

28

-253

• • •

•098

8

4198

•26

72

• • •

28

•187

• • •

•073

6

4151

•18

72

• ■ ■

28

•13

• • •

•051

4

4106

•14

72

• • •

28

•101

« • ■

•089

2

4062

•1

72

• « •

28

•076

• ■ •

•028

1
1

4010

•06

72

■ • •

28

•057

• • •

•022

In Table LV. we have the case of an incompletely
red blind eye, W. The ordinates of luminosity as
measured are given in column III. We have to obtain

298 RESEARCHES IN COLOUR VISION

a factor by which to multiply the numbers in this
column to make it compare with the luminosity of
normal vision given in Table LIV.

Table LV. — Showing W.'s Curves,

I.

IF.

in.

IV.

V.

VL

VIL

standard
Scale No.

X.

Luminosity.

Luminosity

oa+Bs.

from

Col. IV. -V.

RS.
6

(SSN.).
60

6728

^ V V(/f •

Table LIV.

2*5

1-14

• • •

1-14

1-2

58

6521

7-9

3-59

•21

3-38

35

66

6330

20

91

2-25

7-85

7-8

54

6152

42-5

19-32

7-6

11-72

12

52

5996

63

28-66

15*36

13-3

13-8

50

5850

82-5

37-6

26

12-5

12-5

48

6720

92-6

42-08

31-8

10-3

10-8

46

6596

92-5

42-08

32-8

9-3

9-1

44

5481

86

38-7

31-4

7-3

72

42

6373

73

33-2

27-8

6-4

6-8

40

5270

62

28-2

24-2

4

4-3

38

6172

47

21-4

18-6

2-9

2-9

36

6086

32

14-6

12-7

1-9

1-8

34

6002

20

91

7-97

1-1

1

32

4924

12

5-46

4-9

•6

•6

30

4848

8

3-64

3-3

•34

•41

Let us take SSN/s 58 and 46 in the first instance.
The normal luminosities of these SSN/s are 21 and 87
(see Table LIV.), and for W. 7*9 and 925.

From these we can form two equations. Putting z
for the reduction of W.'s total luminosity ordinates and
y for the reduction of those of his red sensation, the
right-hand members of the equations will be formed from
the red sensation luminosities of these two scale numbers
(also given in Table LIV.). The left-hand member of
the equations is the difference between the ordinates of

INCOMPLETE BED AND GREEN BLINDNESS 299

the normal an(i red blind curves at these scale numbers,
which should be equal to the right-hand member —

21- 7-9z = 20%
87- 92-52= 54;ly

From these we find y= 0*829 and 2; = 0-455. Making
X the factor by which the normal red sensation has to be
multiplied in order to give the amount of this sensation
that is present in W/s colour sense, ic = 1 — y, and from
these equations aj= 0*171. That is, when his curve is
multiplied by 0*455, the difference between the ordinates
of his curve and those of the normal give a curve which
is five-sixths of the normal RS. curve.

Taking two other positions, viz. SSN.'s 50 and 44, we
obtain the following equations : —

SSN. 50 . . 100-82*52; = 75y
SSN. 44 . . 75-85:2 =43*3y

From this we obtain y==0*85, 2; = 0*45, aj = 0*15.
Taking the mean of y, we get —

y = 0*835 anda; = 0-165

— that is, W. has only 0*165 (or closely ^) RS.

This number has been used in Table LV. to com-
pare the red sensation curve of the normal with that of
the incompletely blind.

Column I. is the SSN., IL the wave-length (X),
III. the luminosity of the colour blind, IV. the column
III. X 0*455, V. (GS. + BS.) from the Table LIV. ; VI. is
(column IV. - column V.), and column VII. ^ RS. reduced
from Table LIV. It will be seen that after the (GS. +
BS.) have been deducted from the reduced luminosity, we
have a residue which gives (within limits of error of
observation) the same numbers as those given by ^ RS.

300 KESEARCHES IN COLOUR VISION

In this case, then, the incomplete red blind luminosity
curve indicates the truth of the trichromatic theory, and
also of the sensation curves of Table XXXVIII. The
nearer colour blindness is complete, the greater the
necessity for accuracy in the determination of the
luminosity curves.

In the next table is given a determination of a case
of incomplete green bhndness, N.

Table LVI. — N,*8 Curves,

I.

II.

III.

IV.

V.

I

VL

VII.

VIII.

SUndard

Scale No.

(8RN.).

1

1

X.

Original
1 Luminosity
' Readings
byN.

Luminosity
Readings

from
Diagram.

Luminosity

Luminosity ofRS.+BS.

1 XO'82. 1 from Table

LIV.

Column

V.-VI.

showing

N.'sGS.

•13

GS. I
(from
table)
xO-086

60

6728

8-7

8-7

1

7-13

7

• • ■ '

58

6521

25-5

25-5

20-9

20-79

•11

•18 ;

56

6330

68-5

58-5

47-97

47-76

-22

•19 ,

54

; 6152

87-5

89

72-98

72-4

-58

•51

52

5996

100

100

82

80-64

1-36

1 1-32 ;

50

5850

93-5

94

7708

75

2-08

215

48

5720

82-8

82-5

67-65

65-23

2-42

2-73

46

5596

69-6

69-6

57-07

54 29

2-78

2-81

44

5481

57

56-5

46-33

43-69

2-64

269

42

6373

46

45

36-9

34-73

2-17

2-38

40

5270

32*2

34

27-88

25-91

1-97

207

38

5172

23-7

23-7

19-43

17-69

1-84

1-6

36

5085

16

15

12-3

1119

Ml

11

34

5002

8-8

8-5

6-77

6 34

•43

•68

32

4924

4-8

4-7

3-94

3-72

•22

-41

30

4848 '

3-2

3-5

2-87

2-62

•26

•26

28

4776

2-6

2-6

2-13

1-96

•17

•17

26

4707

2-2

2-1

1-72

1-65

•07

-09

24

4639

1-8

1-8

1-43

1-41

•02

•04

22

4578

1-5

1-5

1-23

1-1

•13

•02

20

4617

1-3

1-3

1-06

1

•06

•08

Taking SSN/s 52 and 46, we form the following

INCOMPLETE RED AND GREEN BLINDNESS 301

equations as before ; but from Table LIV. we use the
green sensation luminosity : —

SSN. 52 . . 96-1002 =15-36y
SSN. 46 . . 87- 69-62 = 32-97'

From these we find —

2; = 81, y = 0-906, anda; = 0-094

Other pairs of equations can be formed by, say,
SSN/s 52 and 38 :—

SSN. 52 . . 96 -lOOz =15'36y
SSN. 38 . . 35-9- 2372 = 18-43y

From which we get —

2 = 0-81, y = 0-90, andaj = 0-10

We may take y as 0*90 approximately, which tells us
the green sensation felt is only about one-tenth of the
normal. The green sensation is shown in the table as
0*086 of the normal.

It was not possible to employ this method before the
sensation curves of normal vision had been worked out,
as, unless the composition of the colours in terms of
sensation luminosity is known, y must also remain
unknown.

One more example of the application of the formula
to complete red blindness may be given. In the last
chapter we have the luminosity curve of X. taken direct
in column IX. of the Table L. We can apply the formula
as in the other cases. Taking SSN.'s 50 and 40 —

50 gives 100-252 = 752/
40 „ 50-242 = 25%

' 6S. is 80 small in this, as in the previous cases, that it may be neglected.

302 RESEAKCHES IN COLOUR VISION

Here 2/ = l and z = l. That is, as x = 0, the colour
blindness to red is complete.

Taking SSN/s 52 and 38, we get—

52 . . 96-15-l2; = 80-6y
and 38 . . 36- 18*52; = 17-5y

Here again y = l and z = l, and from this pair the same
deduction is made.

Direct Method of Determining the Colour Sensation

Factor.

We will now give the method of calculating directly
the amount of colour sensation which exists in an incom-
pletely colour blind eye.^ Suppose a person with normal
vision and the person whose colour vision is defective
each make luminosity measures of the same spectrum
colours, the comparison white light in each case being
the same. (The luminosity, it must be remembered, is
measured by alteration in the intensity of the white
beam.) Now the luminosity of the whita light to the
colour blind is less than to the normal eyed by exactly the
amount due to the defect in the red or green sensation.*
Hence, when the colour blind makes an observation, he
is making the comparison with a lower luminosity of
white than doeS the observer with normal vision. If the
white light to each were equally luminous, their readings
would give two curves of such a character that the
difference in ordiriates would be a direct measure of the
defect, as in the previous method. As the white light
is less luminous to the colour blind, we have to find to
what extent the ordinates of his curve have to be altered.

^ The method is adapted also for the completely colour blind.

* The case of blue blindness being exceedingly rare, and the luminoeity of
the blue sensation being so small, we need not consider here this form
of defect.

INCOMPLETE BED AND GKEEN BLINDNESS 303

Let X be the factor giving the amount of his
deficiency in one sensation, and let m, n, and r be
the luminosities of the red, green, and blue sensations
of the ray which is to be measured.

Reverting to Table LIV., the total luminosities
of these three sensations in the whole spectrum of
white light are to normal vision closely as 580, 250,
and 3. It will be seen that the blue luminosity
has but small eflFect, and the red and the green are
nearly as 7 to 3. The total luminosity for the normal
eye is therefore 10. The luminosity of the defective
sensation of the colour blind must be multiplied by a
factor X. Supposing the reading for the normal to be
a, and for the colour blind 6, then we can make an
equation which will contain x. To the red blind n
remains imaltered, and r is negligible, so that we
get the equation in the form —

a{7nx + n) _ b{7x + 3) ^ ,. .

m + n ■" 10 ^^'^

from which x can be determined. When there is
no green sensation in the colour, as when the slit
is at any scale number below SSN. 58, the equation

becomes —

h(7x + 3) ....

For a green blind m remains unaffected, and the
equation (i.) becomes —

a{m + nx) _b{7 + Sx) /... v

m + n 10 ^^"'^

and as there is no green sensation equation,

(ii.) becomes —

b{7 + Sx) ,. X

* {m-\-n) is, of course, the luminosity from Table XX.

304 RESEARCHES IN COLOUR VISION

Supposing x = 0, which is the case when the colour
blindness to red or green is complete, (i.) becomes —

an _^36 i_ lOan
m + n 10 3(m+n)

and (iii.) becomes —

am

7b J
= — or 6 =

m + n 10

lOom

7 {m + n)

(iv.) becomes —

7b , 10a
a=- - or = -=-
10 7

which shows that the readings in the red are larger
for the green blind than for normal vision.

The following observations made by a well-known
man of science, whom we call Z., are given in Table
LVII., and show the application of both methods of
procedure : —

Table LVII. — Showing Z*8 Curves,

1.

Standard
Scale

II.

III.

1

IV.

V.

VI.

VII.

VIII.

X.

i

! LuminoAitT

OfZ.

from

Diagram.

LnmlnoRltT
OfZ.

Lnminofllty
oalculated
from Table

Standard
Scale No.

Origrinal
obaerra-

obeerra-

No.
(88N.).

xO-7.
2-45

LIV.,»S.
being 0*86.

(88N.).

tion.

tton.

60

6728

3-5

2-45

69-6

5

8

58

6521

12

8-4

8*38

67-6

16

25

66

6330

27

18-9

19

55*6

34

50

54

6152

47

32-9

32*9

53*6

53

68

52

5996

62

43-4

43*6

51-6

65

79

50

5850

73

51-1

51*2

■ * •

■ • «

• « *

48

5720

77

53-9

53*9

49*6

74

46

5596

74

50-8

51-7

47*6

76

1

44

5481

67

46-9

46*8

45*6

71

42

5373

67

39 9

39-8

43*6

64

40

5270

47

32-9

33*2

41*6

54

38

5172

35

24*5

24-55

39-6

44

36

5085

24

16-8

16*72

37*6

30

34

5002

15

10-5

10*16

35-6

19

32

4924

8

5-6

617

33*6

11

30

4848

4*5

316

4*06

31*6
29*6

7
4

INCOMPLETE KED AND GREEN BLINDNESS 305

We will ascertain the defect of red sensation by
the first method, and then confirm it by the second
method. From the following table we take the scale
numbers 52 and 46 —

96 - Q2z = 80-6y 87 - 7^z = 54-2y

From this—

y = 0-67 2;=0-68 x^O^ZZ

SSN/s 50 and 66 give—

100-732; = 75y 50-27z = 47-7y

This makes —

y = 0-65 2 = 07 aj = 0-3

From SSN/s 54 and 40—

80 - 47^; = 72-4y 50 - 47z = 25-8y

From this —

y = 0-64 z = 0-72 a; = 0-28

Taking the mean of these factors, we get —

y = 0'65 • 2J = 0-7 aj = 0-3

Here we have the defect in the red sensation is
0*7 ; therefore he must have only 0*3 RS. of normal
vision.

Using formula (i.), at SSN. 59*6, the luminosity of
the normal vision is 8, and of the colour defective 5 —

At another place in the red the readings were 25
and 16 —

At SSN. 55*6 the normal and colour blind readings

u

306 RESEARCHES IN COLOUR VISION

were 50 and 34. In this case m = 52'7 and n = 3'3.
The equation then becomes —

50(52'7a; + 3'3) ^ 3A{7x + 3)

56 ~ 10

This makes —

a; = 0-31

Again, at 53 '6 the two readings were 68 and 34.
The equation is then —

68(74a: + 9-2) ^ 53(7a; + 3)

83-2 "" 10

This gives —

x = 0'S7e

Finally, at 51*6 the readings are 79 and 65. The
equation is —

79(79'5a; + 17'4) ^ 65(7a; + 3)
96-9 10

This makes —

a; = 0-33

The mean of the separate results gives 0*34 as
the factor by which to reduce the normal sensation
for this incompletely red blind. The factor derived
from the first method was 0*3. This example shows
that both methods give the same result within the
limits of error of observation.

The sensation factors from numerous other lumi-
nosity curves, as made from the observations of incom-
pletely colour blind persons, have been worked out,
and so far no case has been met with to which these
methods, founded on the normal colour sensations, as
shown in Table XXXVIII., will not apply. Any small
deviations are readily accounted for by errors in the

INCOMPLETE BED AND GREEN BLINDNESS 307

somewhat difficult measure of luminosity. Whatever
may be the nature of the action on the visual receiving
apparatus, whether it be mechanical or chemical, there
seems to be no reason why similarity in the sensation
curves of the colour blind, compared with those of the
normal curves, should not always be maintained.

A determination of the amount of incomplete
colour blindness, which existed in a recent case, is
now given to show that complete lunvinosity curves
are not required to ascertain the extent of colour
sensation deficiency. The luminosities of only two
points in the spectrum were determined by the colour
blind (Jn.) and the writer. It was found by the
examination that he was incompletely red blind, and
the amount of red sensation deficiency was determined
by the two sets of observations.

At SSN. 34, Jn.'s luminosity was 21, that of A. 45*5

„ 56-7, „ „ 28, „ 43

At SSN. 34, the sensation luminosities from the
table were —

RS. 6S.

and at SSN. 567

6-22

+ 7-98

BS.

OS.

38-45

+

1-55

The following equations were formed to determine
the defect in red sensations : —

14-2-21z = 6-22y 40-28;s = 38-45i/

from which y, the factor of defect, was 0*7, or 0'3
was the amount of his red sensation, and 2, the factor
by which to reduce the luminosity, was 0*47.

Next, using the determinations of the luminosity,

308 RESEARCHES IN COLOUR VISION

the following equations were obtained, where x is the
factor for RS. existing in Jn. s sensation : —

43(38'45.r + 1'55) ^ 28{7x + 3) x = 0-314

40 "10 "

15-5( 6-22a: + 7'98) ^ 21(7,r + 3) ^_o-29
14-2 10

The mean of the two gives 0*3 as the factor, and
agrees with the preceding determination. It is to
be noticed that the blindness must be to the red, for
if we form equations by the first method, supposing
green blindness, with the same numbers we get —

14-2-21z = 7-98y 40-28z = l-55y

This makes y a minus quantity, which is impossible.
Again, with the second method, we should have,
with SSN. 57-6—

43(38-45 + 1-55 0;) _ 28(7 -f3x)
40 *^ 10

where a* is greater than unity.

Caution as to the luminosity method of getting the
factor of deficiency where there is a suspicion that the
macula lutea is very highly or very little pigmented is
here interpolated, and should be read into the results
given in the last chapter. It is safe in such cases to
OoiiLxie the luminosity measures to SSN.'s greater than
42 or 44. With lower SSN.'s the question of pigmenta-
tion may cause a difference in the factors obtained

CHAPTER XXI

COLOUR EQUATIONS FOR THE DETECTION OF

COLOUR BLINDNESS

In this chapter the method of detecting colour blindness,
complete or incomplete, by means of colour equations
made from the spectrum colours will be considered.

Description of White by the Colour Blind.

When a patch of white light is shown to any of the
complete or incomplete colour blind, they recognise it
as their own white ; though not infrequently when they
observe it in contrast with another colour, the latter will
miscall it. But, placed by itself, every person, colour
blind or not, will name it as white. If we place three
slits in the spectrum, one in the red, where it has been
shown that only the red sensation is stimulated, and
another in the green, where the sensation curves tell us
that all three sensations are excited, but the green
mostly, and in excess of the other two, and the third in
the violet, where only the red and the blue sensations
are stimulated, we shall be able, by collecting the rays
on to a patch and altering the apertures of the slits, to
make a mixture which will match a patch of the pure
white when the two patches of light are placed side by
side on a screen. The colour patch apparatus, which
has been described in Chapter IV., p. 38, is perhaps
the simplest apparatus with which to compare the
mixed lights with the white. The normal eye will

309

310 RESEARCHES IN COLOUR VISION

I
I

make his match, which will be exact to him as his wl
If a completely red blind (the eye which sees a shorty
spectrum) is asked if the match is satisfactory to 1
he will say that it is. The completely green blind 1
give the same answer. If the red slit^ be complet
closed, the red blind will see no difference in the mat
for he has no red sensation which can be stimulate
If, however, a partially red blind person be asked if t
normal eye's match is exact, he will say it is not, h
that the composite white is too green. By opening t^
red slit, or closing the green slit gradually, a point wi
be reached in which he says the match is exact. To th
normal eye the match will appear red. If the widths c
the slits be measured, both for the normal and also foi
the colour blind, when the matches to the one and the
other are correct, and if both measure the respective
luminosities of their composite light patches (by opening
or closing the rotating sectors placed in the path of the
white beam which forms the white patch), we have, when
the positions that the slits occupy in the spectrum are
known, a means of calculating the sensation deficiency
in the partially colour blind. If the deficiency in the
colour blind be in the green sensation, the normal eye's
composite white will appear to him as too red. By
opening the green slit gradually, a width of slit will be
found which makes the patch appear to the partially
green blind a match to the white. To the normal eye
it will appear green, more or less pronounced, according
to the degree of lack of response to the stimulation of
the green perceiving apparatus in the colour blind eye.
The slit apertures, and luminosity, of the composite
"white," are measured as before.

^ In this chapter, as in others, the red, green, and violet slits are the
slits through which tht red^ gt^D> and violet rays pass.

DETECTION OF COLOUR BLINDNESS 311

'IC

Formation of Colour Equations.

We will deal with the equations thus formed, which
will be in the form of —

(a) red + h (green) + c (violet) = m (white)

first of all without reference to the numerical value of m,
the sector or annulus reading.

The following are two cases which are dealt with by
this method : —

The three slits were placed at SSN. 59*8 (the position
of the red lithium line), at SSN. 38 '3 (near the green
Mg line), and at SSN. 8 '5 (which is of less wave-length
than G.). A normal eye formed an equation to match
the hue of white —

100 (R) + 40 (G.) + 55 (V.)= White

For convenience in calculation, we can convert the
equation into another, in which G. is 100 —

250 (R) + 100 (G.) + 137 (V.) = White

The comparative luminosities of the rays passing
through equal slits at the three points in the spectrum
which they occupy were R. = 10, G. = 43, V. = 087. In
the red there is only red sensation. In the green there
are red, green, and blue sensations with luminosities of —

RS. GS. BS.

21-18, 21-65, 0-1056

respectively, which make up the luminosity 43. In the
violet ray the luminosity is 0*87 x 135, of which 28 per
cent, is blue sensation and 72 per cent, red sensation.

We will next see how much white the green ray
contains. This is best done by changing the ordinates

312 RESEARCHES IN COLOUR VISION

of the three sensations in the green into ordinates of
the three sensation curves of equal stimulation — that is,
when the areas of the three sensation curves are equal.
In these experiments the source of light was the arc
light with a horizontal carbon for the positive pole (see
Table XL., p. 244). To make the green curve equal to
the red curve, the former had to be multiplied by 2*21 and
the blue curve by 117. The three sensation curves
thus multiplied gave ordinates which when equal make
white. Applying these factors to the components of
the green ray, we get —

RS. GS. White.

G. = 9-46 + 16-55 + 26

The 26 white evidently does not alter the hue of the
mixture which forms white.

The equation, when converted into luminosities,
neglecting the white, becomes —

RS. BS. OS. RS. BS. RS. GS. B.

2500 + 946 + 1655 + 66 + 25 = 3512 + 1655 + 25

R. G. V.

Let us consider the conditions under which a colour
blind person makes a match with a white compared
with one made by normal vision. Suppose we take as
an example a partially green blind as making the
equation.

Firstly, if we call A the luminosity of the white to
the normal, and the luminosity of the white to the green
blind as A', and let x be the factor of the green sensation
deficiency. If the normal equation in sensation lumi-
nosities is —

RS. GS. BS.

a + 6 + c = White

DETECTION OF COLOUR BLINDNESS 313
then the colour blind equation must be —

RS. GS. fiS.

A

{a + hx + c) ,=: Colour blind white

since the only eflfect of the alteration in the white to be
matched is to diminish its intrinsic luminosity.

If we disregard the white luminosity, it is evident
that the equation for the colour blind can be directly
compared with the normal.

A Red Blind Equation examined hy First Method.

In the case of a red blind, his mixture to match his
white was —

100 (R.) + 27 (G.) + 45 (V.) = White

If we make the green 100 as before, it will be seen
that the RS. of the colour blind compared with that of
normal vision will give us the value of x. In the above
equation, doing this, we get —

R. G. V.

370 + 100 + 167

Working this out into the normal luminosity of the
sensations, we get —

RS. RS. GS. RS. BS.

3700 + (946 + 1655) + (104 + 41)
'^ «- -» ^

R- G. V.

or

RS. GS. BS.

4750 + 1655 + 41

To the same amounts of green the amount of red in
the normal is to the red of the colour blind

3512 to 4750

314 RESEARCHES IN COLOUR VISION

that is, the colour blind has only 075 the normal R.
sensation.

The figure obtained by the luminosity method de-
scribed in the last chapter was the same, viz. 0*75 RS.

It is to be observed that the result is obtained by
considering the mixture from a normal eye point of
view.

In regard to the white in the green ray, it is present
to the colour blind as it is to the normal vision, though
it is different in hue, but like the white he matches, and
consequently differs in luminosity, but as it has, as in
the case of the normal eye, no effect on the resulting
hue, it is not taken into account. It has to be re-
membered that to get sensation curves of equal areas
for the colour blind, the factors have to be increased for
the green curve in the case of partial green blindness,
and a factor has also to be introduced for the red curve
in the case of partial red blindness.

Another case is one of green blindness, which will
be the second example of this method of treating the
equation. The observer Y. is a case of interest, as he
has often been quoted as an example of abnormal tri-
chromatic vision.

The measures were taken in the presence of Dr. W.
Watson, F.R.S., with the colour patch apparatus. The
equation of Y. for white was —

98 (R.) + 100 (G.) + 67 (V.)= White

Treating this equation as before, we find that to a
normal eye the equation in luminosities becomes —

RS. GS. BS.

1926 + 1655 + 16

In this case, to get the green sensation present in
the colour blind eye, we must divide Y.'s RS. (red sen-

DETECTION OF COLOUK BLINDNESS 315

satlon) by the normal red sensation or 1926/3562 = 0*54
closely of normal GS. Y/s luminosities at five different
places in the spectrum (see previous chapter for method)
gave a mean value of 0*58 GS. The pigmentation of
Y/s macula lutea was far above the ordinary pigmenta-
tion, and the caution given at the end of Chapter XX.
was observed.

The " white " equations treated this way give trust-
worthy measures of deficiencies where the factors are not
very small.

Second Method of sohnng a Colour Equation,

So long as the factor for the sensation is not below
0*5, it may be followed, but below that point there may
be erroneous estimates derived from the calculation.
The normal eye cannot detect within 2 per cent, of
excess of a colour matched to a white, and guard had
to be taken against this in forming colour equations, to
ascertain the spectrum colour sensation curves for the
normal eye. There is reason to believe that for small
sensation factors a much larger quantity of colour may
be added to white, and of white to the colour, than can
be added by the normal eye without detection. It has
already been pointed out that, to a completely red blind,
the match to the normal eye is satisfactory, although it
is just as satisfactory to him if the red slit be closed.
Indeed, any amount of red may be added to his white
without altering the match. We can understand that,
with an eye which only has, say, 005 RS., an almost
equal amount of red might be added to the white and
not be perceived. As the factor increases, the amount
of white that can be added to the red, or of red to the
white, without altering the hue, will be less — and so also

316 RESEARCHES IN COLOUR VISION

with the green sensation. It seems that the ordinates
of a curve that may represent the amounts that can be
added may probably be the ordinates of an hyperbola.

Whatever may be the reason of the want of percep-
tion of the added colour, we know that the want exists,
and the second method of treating the equation gets
over any difficulty on this account. The method is a
combination of the first method with that of the lumi-
nosity method. If when the white is matched in hue
by the colour blind, he is also required to make a deter-
mination of the luminosity of his composite white, and if
the normal eye also takes a measure of the luminosity of
the colour blind composite white, or takes a measure
of the luminosity of his own composite white, there are
sufficient data with which to calculate the sensation
deficiency. It should be noticed that the luminosity of
a composite white against a pure white is very easily
measured. There is no difficulty in the observation,
though it may exist to some observers when the lumi-
nosity of a colour against white has to be determined.

We will suppose that the following equation has
been made by a green blind : —

a (R.) + b (G.) + c (V.) =p of sector to the colour blind

and that to the normal eye it has a luminosity of m. It
is only necessary to take into account the luminosities of
the red and green sensations, since those of the blue
sensation are very small compared with them.

Let us turn the colours into sensation luminosities,
this time not calculating out the white in the green ray,
and the equation becomes to the normal eye —

RS. GS.

a + 6 = km

DETECTION OF COLOUR BLINDNESS 317

A being the factor which makes m = (a + 6). Using A. for
the green blind equation, we have /ip, but to the colour
blind p is dependent on the area of his total luminosity
curve, which is smaller than the area of the normal
luminosity curve of the spectrum.

Let A be the area of the normal luminosity curve
(Table XL.), and A' the area of the colour blind lumi-
nosity curve.

To make 2^ balance the composite white to the normal
eye, the left-hand members of the equation must, as on
p. 313, be multiplied by A. A', and calling x the factor
of the sensation deficiency for the colour blind, we get
for green blindness —

A'

or a-^l)x = hp-^

RS. GS.

If the value of A be 10, i.e. (6-8 + 3-2)'—

RS. GS.

A' is 6-8 + 3-2:c
A',A = (0-68+0-32a;)

^ _0 '68^/>~ a
h-O'S'Zhp

If the deficiency were in the red sensation—

' _Q'68 ^p-6

The value of h may be determined, we said before,
by the normal eye measuring his composite white against
the same white patch which the colour blind matched.

^ These numbers are derived from the himinosity sensation (R. and Q.)
I'urves of the light used in these measures, Table XL., p. 844.

318 RESEARCHES IX COLOUR VISION

It will be noticed that x is determined regardless of
the true amounts of RS. and GS. on the left-hand side
of the equation.

The following is an example of what may be called
a glaring case of an untrue equation being formed by a
nearly completely red blind person (S.). The mean of
two of his equations was —

30 (R.)+ 1675 (G.) + 12-75 (V.) = 27-2° of sector in white

We may neglect the luminosity of the blue sensation
and use only the red and green.

Converting the above into luminosities of RS. and
GS. (in this instance not taking away the white which
is in the green ray, as all its components of red and
green sensations are required), viz. —

RS. 68.

21-18 and 21-65 (see ante)

and having found from a normal vision equation that
^ = 41, we get —

RS. RS. GS.

300 + (355 + 365) = 27-2 x41(0-68a; + 0-32)
G.

RS. GS. RS. GS.

or 655a; + 365 = 758a; + 357

From this we get —

.r = 008 nearly

or S. possesses about 0*1 of the normal RS.

Using the first method of treating the equation,
he would have been supposed to have 0*8 RS. His
RS., calculated by the luminosity method given in
Chapter XX., was 0-1 closely.

DETECTION OF COLOUK BLINDNESS 319

A case of green blindness (Wn.), which gave a fairly
large deficiency by the luminosity method, is now given.
His equation to white was —

30 (K) + 32 (G.) + 39 (V.) = 23 White

At the same time, and using the same comparison
white beam, a person having normal vision found an
equation which gave a factor h for the white of 67.

Applying this factor to Wn/s equation, we get as
the luminosity equation —

RS. 6S.

978 + 6910:= 1048 + 4930;

or = 0-35 of normal GS.

His factor of GS., obtained by the luminosity method,
was about 0*33.

If we treat Wn/s equation by the first method, we
get a factor of 0*54.

These two cases confirm what has been said as to
non-recognition of white or colour when added above
the 2 per cent, limit.

It must be remembered, in accounting for the lack of
accuracy in mixing the colours to form white, that to the
normal eye the white of the largely deficient green blind
is a slightly pale purple, and that of the largely deficient
red blind a slightly pale sea green.

[In the examples given, the position of the green slit
may seem not to be the best one to use, as this ray,
besides the white, contains both green and red sensa-
tions; but for general purposes it is a good one. The
ideal position is that the ray which passes through the
slit should only be composed of white and green sensa-
tions. This position on the standard scale with the arc
light and horizontal carbon is close to SSN. 36, but it

320 RESEARCHES IN COLOUR VISION

must be remembered that this position is one in which the
rays are largely absorbed in most instances by the yellow
spot.

When the distance of the eye from the screen is kept
absolutely constant, it is preferable that the ray should
contain white, green, and a trace of blue sensations,
rather than white, green, and red sensations, as the
latter imposes a limit on the green sensation factor. In
the position SSN. 38'3, which the slit has occupied in
the above examples, the limit of the factor is about
0*26 GS. For the red deficiency there is no limit when
using that position.]

Examples only of incomplete colour blindness have
been given. When the colour blindness is complete, only
two slits need be opened. The third (red or green) may
be opened to any extent, but the last method will show
the ''completeness" of the sensation's deficiency.

CHAPTER XXII

MATCHING A PURE COLOUR BY A MIXTURE OF TWO
COLOURS, AND A MIXED COLOUR MATCHED BY
ONE PURE COLOUR.

A FAVOURITE plan in Germany for a semi-quantitative
measure of colour sensation deficiency is that which
originated with Lord Rayleigh. This method is one of
mixing red and green to match the sodium D light of the
spectrum. There are special instruments extant for this
purpose, and note is directed to be made of the quantities
and intensities of each colour which are required to give
a match to this light. There are, however, no directions
given by which the factor of deficiency is to be ascer-
tained, though it would be easy to give them when the
positions of the red and the green in the spectrum are
known.

Matching of the * * Z> " Light.

If we place two slits in the colour patch appar9,tus in
the same positions that we have already used in the red
and the green, we can make an approximation to the
deficiency by the match made of the D light. The
match made will be of the same hue as the D light when
a little white is added, for there will be white in the
mixed colours. In Chapter XVII. it is shown that from
the scarlet to the greenish yellow in the spectrum the
addition of white to a colour will make its hue yellower,
and from the blue-green to the green the same ** yellow-
ing " of the hue would be apparent. •

322 RESEARCHES IN COLOUR VISION

In matching the D light with a green (every green
contains white) and a pure red, the true proportion of
RS. and GS. in the match will not be quite identical
with those in the D light itself. If the colour to be
matched be at SSN. 48'7 of the standard scide, which is
where the red and green sensation curves of the arc light
spectrum (of equal areas) cut, this would not occur, since
at that point no change in hue is found when white light
is added to it.

If, however, a light such as the paraffin light is
employed as the source for the spectrum, the red and
green curves of equal areas will cut very close to D in
the spectrum, and the white light existing in the green
ray, when calculated out (as has been done for the arc
light), will be very nearly the hue of the D light, so that
there will be no shifting of hue. It is necessary to men-
tion this, as, if the match is to be used for ascertaining
colour sensation deficiency, the sensation curves for the
light source used must be employed in the calculations.

A gauge of accuracy of measurement is the closeness
with which the mixture of red and green made by a
normal eye shall give the hue and the proportion of
sensations existing in the D light.

The writer's mean equation for the D light, with the
slits in the same position as before, is —

447 (R.) + 100 (GO-D light

This, when worked out with luminosities, gives a per-
centage value of —

RS. GS.

n -I- 23

as contained in the mixture, neglecting the white. This
is very slightly (O'S*) less red than is contained in the

[,)V MATCHING A PURE COLOUR 323

v;:- D light, and is probably to be accounted for by the white
existing in the green ray.

There is in these D equations, as in the equations for
white light, the same possibility of their failure when
the sensation factor of deficiency is small owing to the
non-perception of added colour, but if the luminosity of
the D light (or other selected ray) be measured against
the mixed colours, the difficulty, as before, vanishes.^

Matchiny the colour of Chromate Potassium mth

a Single Ray.^

In Chapters XVIII. and XIX. several methods have
been described for ascertaining quantitatively the amount
of green or red sensation which exists in the incomplete
green or red blind eye as compared with the normal
eye. In Table XXXVIII. , at page 239, is to be found
the percentage composition of the tabulated rays of the
spectrum, and Table XXXIX., p. 242, gives the amount
of white (where there is any) which exists in these
several rays. For reasons given later, attention must
again be called to the fact that if the colour at this point
is mixed with blue at SSN. 23, by the proper adjustment
of width of slits a match can be made of the white light
which goes to form the spectrum. (Whatever the source
of light, the curves of equal areas must be calculated
for it, as the point of intersection varies according to
the light employed.)

Looking at Table LIV., p. 297, it will be seen that
from SSN. 50 to the extreme red no measurable quantity
of blue is to be found. If the beam of (say) the arc
light has to pass through a cell containing a saturated
solution of potassium chromate of about J inch in thick-
ness, the light will become yellow with very little blue

» See Chapter XVII. « See Paper No. 26.

324 RESEARCHES IN COLOUR VISION

in its spectrum. If in the colour patch apparatus a slit
be caused to traverse the spectrum, a position will be
found where the ray passing through it exactly matches
the hue of the white light after transmission through the
chromate solution. This will be at SSN. 496 (X 5830)
to the normal eye. This, like other rays, contains a
fixed ratio of green to red sensation, but no measurable
blue, and therefore no white which could alter the hue.
Making the red sensation unity, it will be found that
the green sensation is '385 at SSN. 49*6.

The following table gives the ratios of green to red
for the standard spectrum scale, making red unity, as
also the wave-length scale for the like ratios : —

Table LVIII.

SSN.

GS.

X.

GS.

GS. corrected
from Diagram.

56

•047

6300

-050

•050

54

•105

6200

■080

•080

52

•187

6100

•127

•127

50

•333

6000

•185

•185

48

•475

5900

•280

•280

46

•603

5800

•390

•385

44

•717

6700

•500

•490

42

•830

5600

•600

•595

40

•934

5500

•700

•700

38

1-05

5400

•805

•805

36

M6

5300

•910

•910

34

1-26

5200

1*015

1^015

32

1-33

5100

1135

1120

30

1-29

5000

1-260

1-225

28

114

4900

1-340

1330

26

•82

From these tables diagrams on a large scale can be
drawn from which the ratios of red to green can be read
off for any scale number (or wave-length). Fig. 9 1 gives
such a diagram on a small scale.

If an incompletely green blind makes a match, the

MATCHING A PURE COLOUR 325

slit would have to be moved towards the red. When
he considers the match correct, the scale number of the
ray is read off and a reference is made to the diagram.
Thua, suppose that the mean reading of the match were
52, the amount of GS. to RS. to the normal eye would
be 0*187. By dividing this number by '385 (the number

Fio. 91. — Figure sbowlug ratio of green to red sensation.

corresponding to the match for the normal eye), we get
very closely 0*5, and this would be the amount of green
sensation (compared with 1 for the normal eye) that the
green blind possesses. Again, if we have the match by
an incompletely colour blind at SSN. 46, we know at
once he is incompletely red blind, as that SSN. contains

326 RESEARCHES IN COLOUR VISION

•603 of GS. 1 of RS. Dividing 385 by 603, we make

the amount of RS. which he possesses as '638.

An inspection of Fig. 91 shows that the maximum

ratio of green sensation is near SSN. 30 when it is about

1*32. As the normal match has '385, and as this has to

be divided by the incompletely red blind person's ratio,

it shows that no smaller factor of red sensation can be

'385
found than ~- or -29 RS. * For the green blind the

X 'O^

smaller factor can be found, but the test is especially
useful for large factors.

One example of the accuracy and delicacy of the
test is now given. The normal eye made a match at
SSN. 496, and an incompletely colour blind at SSN. 41'5.
The former, as before, has a ratio of '385, and the latter of
'855 red to green sensation. This gave a factor of '45
for the green blind's GS.

The same person was tested by the luminosity
method described in Chapter XVIII., which also made
him have '45 GS. From his colour equations his factor
was -37. The mean of the three values derived for his
factor is '42 but '45 is most likely to be right.

In making these tests, the luminosity of the white
beam passing through the chroraate is first made to be
about the same as that of the light coming through the
slit. Four matches of colour are sufficient, two by
reaching the match from the red side, and another two
from the blue side of a first approximate match. A
mean of the four readings is taken as being the position
of a correct match, though not unfrequently all four are
the same.

It may be advisable to indicate how the amount

* When bichromate of potassium is substituted for the chromate, smaller
factors can be measured.

MATCHING A PURE COLOUR 327

of displacement, if any be possible, of one or other of the
green and red sensation curves can be determined.

At p. 323 it was pointed out that when a slit was
' placed in the ray where the two green and red sensa-
tion curves of equal area cut, the addition of pure blue
enabled a match to be made with the white of the light
which formed the spectrum.

Let aa, hh be two portions of the green and red
sensation curves respectively which cat at and having

an ordinate OC, then a slit placed at C in the spectrum
will allow a ray to pass, which with the blue of SSN. 23
will match white. This holds good also for the colour
blind, since the curves under consideration are " eqiial
area" curves. The white they would match would of
course be the "colour blind white." If the green curve
were shifted to the left, the curves would no longer cut
at O but at O', and the slit would have to be placed at
C before white with the blue would be produced. The
same occurs also should the red curve be shifted. By

328 RESEARCHES IN COLOUR VISION

I:

making observations such as this entails, any shift can
be noted. Should such an alteration in the position of
the intersection of the two curves take place, the differ- \

ence in position of the slit placed in the yellow must \
be added to, or subtracted from, the position of the J
normal match for the eye when the colour of the
chromate is used for matching. If there be colour
blindness, this corrected position, if ever found, might
have to be used for comparison.

It may be remarked that by the luminosity method
of ascertaining the factor, a non-normal sensation would
be closely the mean between the factors shown on the
red side of the maximum luminosity, which might differ
slightly from those obtained on the blue side.

CHAPTER XXIII

MEASUREMENT OF GREEN OR RED SENSATION
DEFICIENCY BY MEANS OF COLOUR DISCS

The methods of ascertaining the amount of colour sensa-
tion deficiency in the colour blind have so far depended
on measurements made in the spectrum itself, but atten-
tion must be called to a method which is independent
of a spectrum apparatus. It is true that its accuracy
in the first instance depends upon measurements made
in the spectrum ; but when once made, a colour sensa-
tion deficiency (within limits) can be determined without
further reference to it. We mean by colour disc equations.
Given three discs of equal diameter (say 4 in.), capable of
interlacing and of being rotated, one of which is painted
with a red pigment, another with a green pigment,
and the third with a blue pigment : by altering the
angles of the interlacing discs, a grey can be formed
on their rotation,^ and this can be matched by a white
and a black disc of, say, 6 in. diameter, also rotating on
the same spindle. Of course there is nothing new in this
method, but the method of treating the equations given
by the colour discs will be found new in some details.
Colour discs can be used in any lights but to be really
useful for calculation the kind of light should be known.
The colours of the discs themselves are the only
part of the apparatus which requires careful measure-
ment, and this must be done in the spectrum. The
composition of the colours must be ascertained in terms

* See Chapters XI. and XVI.

820

330 RESEARCHES IN COLOUR VISION

of the three colour sensations, and the luminosity of the
colours must also be known. The former and the latter
will both vary according to the nature of the light in
which they are viewed.

Spectrum Composition of the Pigments.

We may proceed in ascertaining the composition of
the pigment colours by the method given in Chapter
XVI.

The compositions of the pigments are there given for
the light of the electric arc ; but when the luminosity
curve of the spectrum of any other light is known, the
sensation luminosities in the pigment colours can be at
once calculated from the table at p. 239.

Let the amount of each ray which is reflected from
the pigmented surface be measured. Such a method
also gives the luminosities in terms of the total white
light used to form the spectrum. This is an exact
method, but a somewhat long one, and perhaps it tells
more about the pigment than is necessary to know for
the purpose that is in view. All we require to know,
as said before, is the composition (in sensation luminosity)
and the total luminosity. The former we can arrive at
in a very simple manner. Let us place a square piece of
the pigmented paper in the colour patch apparatus, and
side by side with it an equal square of a white surface.
Let the pigment patch be illuminated by the light in
which the discs are to be used, say, gaslight, incandescent
light, &c. (daylight is out of the question, as it is so vari-
able in quality), whilst the other is illuminated by the
arc light coming through the three slits in the spectrum,
as has already been described. By placing a rod in the
path of the beams, the two illuminations may be separated,

MEASUKEMENT BY COLOUR DISCS 331

but can be caused to touch one another. All we have to
do is to match the colour of the pigment, as seen in the
light by which it is illuminated, with the mixture of
the rays coming through two or three of the slits.
The light itself is also evaluated by making both patches
of zinc white, one being illuminated by the light to be
employed. Having done this, the width of slits must
be measured as before described. When converted into
luminosities, and the luminosities into the respective
sensations existing in the rays, the relative amounts of
the sensations stimulated by the pigments and by naked
light can be calculated. By making the patches equally
bright, the relative luminosities of the pigments com-
pared with that of the light illuminating the white can
also be ascertained with great exactitude if the pig-
mented paper is removed and a second square of white
paper is substituted for it. The sensation values of the
three coloured discs for the light in which they are to
be viewed will now be known, as also the luminosity.

An example will show that both methods of ascer-
taining the sensation luminosity values of the light and
pigments give within small limits the same values.

The comparison light was the reflected arc light as
used in the colour patch apparatus, see p. 39, with a cell
of potassium chromate placed in the beam. The absorp-
tion of the chromate solution was measured and con-
verted into luminosities by the method given on p. 76.
The red and green sensations were calculated. The
intensity of the reflected light from the pigments was
measured, as given on p. 78, and from their luminosity
curves and percentage sensation curves (Table XXXVIII.)
the luminosities in red and green sensations were cal-
culated.

The colour of the light passing through the chromate

332 RESEARCHES IN COLOUR VISION

solution on to the white and on to the pigment patches
was matched by a mixture of the rays passing through
the red and green slits in the spectrum, and the sensations
were calculated as before, with the following results : —

Light. First Method. Second Method.

RS. GS. Ra GS.

Chromate on white . 708 + 292 711 +284

Chromate on green pig-
ment . 62-7 + 37-3 62-6 + 37-4

Chromate on red pig-
ment . . 85-3 + 14-7 84-7 + 15-3

Using the Colour Discs.

To use the discs to give true equations, the illumina-
tion must be that of the same kind of light as that in
which their sensation values have been determined. It
will not do, for instance, to use the values obtained for
the arc light in daylight or in gaslight. If an incan-
descent light (say) is used for the illumination of the
pigment during measurement, the discs must be rotated
in the same light. Stress is laid on this, as it is not
uncommon for those using colour discs to be lax as to
the light they use.

The three discs are placed on the spindle of the
whirling apparatus (a small motor is handy for the
purpose) with the interlaced black and white discs behind
them. The coloured discs are altered till a grey is
obtained which matches the grey of the rotating black
and white discs.* The angular apertures of the exposed

^ It is well that the matches should be uiade with the light falling per-
pendicnlarly on the discs and the observer being as nearly as possible facing
them.

■ V B • li.a

MEASUKEMENT BY COLOUR DISCS 333

parts of the several discs are all measured and the values
recorded as —

a red + h green + c blue = m white + (360 — m) black

The amount of white reflected from the black is
measured, and if n be the factor the white becomes
m + (360~m)n.

It is essential in some cases that both the greys
should be of exactly the same brightness. (It need
scarcely be said they should be identical in hue.) Every-
thing depends, for a true determination of the amount of
colour blindness, on the true matches being made.

It may here be emphasized that both luminosity and
sensation composition will vary in every light, so that
exactitude of match in any light but that in which the
measurements have been made is labour thrown away.

Examples of Colour Disc Equations.

We will now give examples of the mode in which
the equations should be treated, and this will be similar
to that of the spectrum equations in Chapter XXIII.
The light in which the rotation of the discs was made
is the naked arc light, and all the measures were made
in that light.

The following is the equation made with the discs : —

126 red + 144 green + 90 blue = 79 white (i.)

and the black reflected just 5 per cent, of white light,
so that the equation on the right-hand side becomes —

79 + 281x0-05 or 93°
The composition of the vermilion red was found to

334 RESEARCHES IN COLOUR VISION

be (in terms of the luminosity of the whole spectrum,
and which equalled in area 866 on an empyric scale) —

RS. G& White.

142-5 +165 + 53 (ii.)

the emerald green was —

31-8 +60 +263 (iii.)

and the blue was 234+ 1-56 +.342 (iv.)

Multiplying the equation (i.) by the appropriate
factors in (ii.), (iii.), and (iv.), and dividing by 360**,
we get —

RS. GS. BS. White.

49-9+ 5-8 + + 18-55
127+24-0 + +105-2
0-6 + 0-4+ 8-55
62-6 + 30-4 + 0-4 + 132-3

Dividing this equation by 866, we get the sensation
luminosities for the mixed colom:s —

RS. GS. BS. White.

0-0725 + 0-0351 + 0-0005 + 0-1527 = 0-2608

The ratio of RS. to GS. is 67 to 33, which is closely
that obtained from the spectrum equation, so that the
above equation derived from the discs may be taken as
the normal vision equation.

We do not need to refer to the right-hand member
of the equation, but if we take it as 93 the luminosity
of the white exposed is 93/360 of 1, 0*259.

It will be seen that the luminosities agree to within
the third place of decimals, as the left-hand member
comes out at 0-2608.

When a colour blind person is tried in the same light,
his equation is —

210 red + 100 green + 50 blue = 77

MEASUREMENT BY COLOUR DISCS 335

Taking the luminosities of the red, green, and blue
as before, we get, when multiplying them by the equa-
tion numbers (ii.), (iii.)j ai^d (iv.)^360 —

RS. GS. BS. White.

91-94 + 26-G + 0-22 + 1087
Dividing by 866, as before, we get —

Total
RS. GS. BS. White. luminosity.

01062 + 0-0307 - 0-00025 + 0*1253 = 02624

RS. is to GS. as 114*2 to 33, the normal equation
being as 67 to 33.

The degi'ee of red blindness is given by 67/114*2, or
0-58 RS.

We may now examine the right member of the
equation, which is the white in the outer two discs
of black and white. It is 11 ^ and, with the light re-
flected from the black, becomes 91, and 91/360 = 0-2527.

We may now subtract the white of the left-hand
member from it, and we get the following equation
left :—

RS. GS. (RB., white).

01062x-t-0-0307 = 0-1274

where x is the RS. factor.

As in the second method of using the spectrum
equations for the colour blind, we multiply 0*1274 by
(67ic + 33), as (67 + 33) is the normal relation of RS.
toGS.

This worked out gives x = 0*54 RS. Another colour
blind makes the same equation match with 72 white.

Proceeding in the same manner, we get —

RS. GS. RS. GS.

0-1062ic + 00307 = 0-0768x + 0*0379
a; =0-24 of the normal luminosity

336 RESEARCHES IN COLOUR VISION

We see, then, that where there is a deficiency in the
mixtures due to causes ah^eady pointed out, the degree
of colour blindness can still be calculated, always suppos-
ing that the black and white mixture is to the observer
a perfect match to the inner grey given by the discs.

The question of other illumination need not be
entered into by examples. They would be carried out
in exactly the same manner as that indicated. The use
of colour discs to form equations, as before said, has long
been known, but the method of using the equations in
the manner indicated above is apparently novel.

For general use in forming equations for the colour
blind, a yellow light is one that commends itself to the
writer. When the light is white, the amount of blue,
which is not much more luminous than the black that
has to be mixed with the red and green in the inner
disc, is so great that the grey produced on rotation is
dark. On the other hand, if the white arc light, or,
indeed, any other light, is transmitted through a chromate
solution as given before, no blue in the inner disc is
required to match the outer grey disc. Both the disc
and the outside ring are fairly bright, and the matches
become easy.

One point must be mentioned which to some extent
prevents the disc equations being as useful as the
spectrum equations. In the red pigment used, always
a certain amount of green sensation is also excited, and
in the green a certain quantity of red sensation is
excited. It follows that even with an all-red centre the
factor of the partially red blind who can make a good
match with the outside grey is limited, and any addition
of the green disc will not diminish but only increase the
proportion of red to green sensation in the centre disc.
The same applies to the green blind with the green disc.

MEASUKEMENT BY COLOUK DISCS 337

A completely red or green blind will, however, match
both discs with a grey, since the green or the red sensa-
tion will be completely absent to them.

A simple method of arriving at the sector angles at
which the different degrees of colour blindness will
match a grey, is to place a **chromate" light in the
reflected beam of the colour patch apparatus (or in any
light which is to be used to illuminate the discs). The
discs are then rotated at the colour patch screen, with a
white patch alongside. This can be effected as usual
with a rod placed in the two beams. The colour of the
white is first illuminated by the ** chromate" light and a
match made with it (see p. 324). The plain red and
the plain green are then matched with the single spec-
trum colour. The red and green discs are then inter-
laced say, with 30 red and 330 green showing. When
the compound disc is rotated, a single colour will match
the mixed colour which is noted. By taking more red,
another match will be found, and so on. These single
spectral colour matches are then applied to Table LVIII. ,
and the degree of colour blindness which they indicate
calculated. The different sectors of red and green can
then be readily ascertained for any required degree of
partial colour blindness.^

' There will, as stated above, be a limit to the factor.

CHAPTER XXIV

SOME CASES OF UNCOMMON COLOUR

BLINDNESS

In this chapter a few recorded cases of colour blindness
diflTerent to those ordinarily found are given, and to
most the method of ascertaining the amount of colour
blindness has been applied.

Cases of Monochromatic Vision.

The first is a type of colour blindness in which all
sense of colour is lost. Reds, greens, blues, yellows,
and, in fact, all colours can be matched with one
another, they all being different shades of grey to
this type of colour blind. This type of colour blindness
is congenital ; at least it was said to be so in the few
cases examined. Both eyes were similarlkr affected.
It is usually supposed that this monochromatic vision
is due to some form of disease, but it seemg. to be, if
not hereditary, at all events found in the same genera-
tion of a family. Two cases, which we call P. and Q.,
are examples (Fig. 91). They were brothers, and had
identical lack of colour perception. Their luminosity
curves are valuable, as they practically coincide with
the luminosity curve of a feeble spectrum, which is
given in Chapter VIII., showing that in the feeble spec-
trum the luminosity is principally due to the fundamental
sensation of light in normal vision. The figure shows
the luminosity of the spectrum to P., and also the

888

UNCOMMON COLOUR BLINDNESS 339

extinction of light measures that he made. The in-
tensity to the normal eye of the D light was equal to
one candle at 1 foot distance from the screen. In
the table which follows, the readings in column IV. are
in millionths of the original intensity. In column V.
we have P.'s persistency curve, with a maximum of 100.
Column VL is the luminosity curve taken direct. It
will be noticed in comparing these two curves, that the
readings in the blue-green, blue, and violet are smaller
in the luminosity curve than in the persistency curve.
This no doubt is due to the fact that the luminosity curve
was taken when the images of the patches fell on the
yellow spot, whilst in the extinction curve the eye was
allowed to wander when looking for extinctions, and
is to be collated with the results given in Tables XVII.
and XVIIL for the normal eye.

If we multiply column VI. by IV., we get the
value of the extinction when the rays are made of
equal luminosity. As far as the blue-green they have
equal values (about), but diminish from the blue-green
to the violet, due to the cause to which attention has
just been called. It thus appears that P. has only
one sensation, that of light, since the extinction value
of every ray when of equal luminosity is, more than
probably, the same.^ From other evidence it appears
that P. and Q/s sensation of luminosity for total white

is about ^V ^f *1^® normal.

As cases of monochromatic vision are rare, the follow-
ing one is also put on record.^ The patient, whom we will
call K. B., was kindly brought by Mr. Parker. He was

' If the method of ascertaining the amount of colour blindness for red
or green be applied to these curves, they will be found to give impossible
factors.

• Paper No. 16.

340 RESEARCHES IN COLOUR VISION

aged twenty-five at the time when examined for colour
vision. The notes of his case were as follow : — " Vision

*^^Y h ^?*^°*>™ ; has always been colour blind. Has
q *^'^zontal nystagmus ; probably an absolute cen-

UNCOMMON COLOUR BLINDNESS 341

tral scotoma. He is always * day-blind/ His vision for
right and left eyes is 6/60. He is not night-blind. His
fields are nearly, but not quite, full for white. He shows
no definite changes in his eyes."

Tablb LIX. — P. ^8 Luminosity and Extinction Curves,

I.

II.*

III.

IV.

V.

VI.

Mean reading

Adopted

Persistency

curve

680

S<vile No.

Wave-
lengths.

of extinction in

millionths of

original

reading in

millionths of

original

P.'s
luminosity

All W^V A

ad. reading.

curve.

luminosity.

luminosity.

52

5996

68

68

10

7

60

5850

35

35

19-4

19

48

6720

17

17

40

39

46

5596

10-2

10

68

65

45

5538

9-3

9

76

76

44

5481

8

81

84

90

42

5373

72

7-2

94-5

98

40

5270

6-7

6-8

100

99

38

5172

7-2

7

97

97-5

36

5085

805

7-7

90

90

34

5002

8-05

8-4

81

80

32

4924

9-9

9-8

69

65

30

4848

13-2

12-5

54

50

28

4776

13-9

15

45-3

36

27

4742

16-8

17

40

31-5

26

4707

21-6

20-6

32

26-5

24

4639

30

27

25

19-5

22

4578

36

35

19

14

20

4517

42

45

15-5

10

16

4404

79

79

8-5

5-5

10

4245

180

190

3-6

2-5

6

4151

270

270

27

• • •

In taking his luminosity curve, he matched all colours
with white with the same facility as if they were white,
though he was not a good observer at first. The follow-
ing table gives the luminosity of the spectrum to him,
and for the convenience of reference P.'s curve of
luminosity is given for comparison : —

342 RESEARCHES IN COLOUR VISION

Table LX.

Scale of

spectrum

(prismatic).

56
54
52
50
48
46
44
42
40
38
36
34

Scale of

K. B.'8

lumiausity.

P.'«
luminosity.

spectrum
(prismatic).

K. B.'8
luminosity.

P.'a
Inminoaity.

2-6

32

61-5

65

9

—

30

43

50

16

7

28

37

36

27-5

19

26

30

26-6

42-5

39

24

24

19-5

61

65

22

18-5

14

82-5

85

20

145

10

96

98

18 •

11-5

""""

100

99

16

9

5-5

95-6

97-6

14

7

87-5

90

12

5

75

80

10

3

2-5

It will be remarked that the maximum of each curve is
about scale number 40, or close to E. On the red side
of the maximum the curves do not absolutely agree.
K. B/s observations were first made in the red and
green, and his readings at first were not very close, and
a mean had to be taken. As the colours he had measured
went towards the blue, his measures were much more
accordant, as he had become accustomed to the methods
employed. The slight divergence on the left-hand side
of the curve from that of P. is probably due to the
colouring matter in his yellow spot. Attention must
be again called to the fact that both P.'s and K. B.'s
curves are practically identical with those obtained by
the normal eye when it measures a spectrum of very
feeble luminosity, and also agree with the results ob-
tained by measuring the diminution of each ray when it
first becomes invisible, and making a curve of the
reciprocals of the numbers, taking the highest point
of it as 100.

UNCOMMON COLOUR BLINDNESS 343

A Case of Colour Blindness and Lack of Pigment.

A case * is now given in which the absorption by the
yellow spot pigment seems to be absent, coupled with a
considerable amount of red blindness. This was a re-
markable case, which Mr. Nettleship mentioned.^ He
had stated that this lady, N. W., mistook blue for red,
and it was with some curiosity that this case was
examined. Her first examination was as to colour sense
with the spectrum colours, a patch of monochromatic
light being placed in juxtaposition with an equal patch of
white light. At 62-5 (X 6890) of the scale the red of the
spectrum disappeared. As the slit moved along the spec-
trum, and the white was approximately reduced to equal
luminosity, she described all the red as grey, and of the
same colour as the white until 53'5 (X 6110), and after
this point she said the colour was brownish compared
with the white. The colour continued of this hue to her
till 48 on the scale (X 5720), when she said the colour
was neither brown nor green, but both. From 48 on
the scale she described the colour as green till it sharply
ended at 31*5 (X 4905). In the blue she again began to
see grey ; the grey at this end of the spectrum, and also
of the white patch, she called brownish grey. [This
name must evidently have been a mental distinction, as
she described the red end and the white as grey only,
and not brown-grey ; and, indeed, she was tried again
over that part of the spectrum, and adhered to the
previous naming. It would appear to be due to the
low luminosity which made the grey appear brownish
to her, and not to any actual difference in hue.]

» Paper No. 17.

'To a Committee of the Royal Society on Colour Vision which was
sitting at that time.

344 RESEARCHES IN COLOUR VISION

Her curve of luminosity in the spectrum was next
taken, and her readings are given in the table. The
curve is shown in Fig. 94. The shaded band beneath
it applies to her curve.

Table LXI. — Showing N- W.'» Ou)D€S.

I.

I

Standard
I Scale
Nos.

(SSN.).

62^5
i 60

68
56
54
52
50

48
46
44
42
40

38
36
32
31
25

20

16

10

H.

III.

N. W/«

X.

Lumi-

nosity.

7020

6728

3

6521

10

6330

30

6152

52

6996

70

6860

81

5720

87

6596

90

6481

88

5373

72

6270

62-6

6172

46

6086

23

6002

12-6

4885

10

4675

6

4517

3

4377

2-5

4245

1-6

4010

•2

IV.

v.

vr.

N. W.'s Luminomtv from
Lumi- Table LIV.,

nofiity RS. being 0'25
XO-S. of Normal.

N. W. named the prismatic oolourn
against white.

1-5

1-75

Both grey

6

5-4

16 !

142

26

25-7

36

356

40-5

43-7

M

Colour brownish, white grey

43-5

45

44

36

31-2

23

11-5

6-26

5

2-5

48-1
46-3
41^8
36-5
30-5

22-9
12-4
5-67
4-67
1-41

»»

w

*)

»J

It

»>

Colour brownish-green, white grey
Colour green, white grey

»»
»»

»♦

! (Colour brownish-grey, white
\ brownish-green

1-5

•52

»»

>»

1-25

•28

yy

ft

•76

•16

ft

ft

•1

•014

ty

»>

•»
f*

ft

The table shows that the readings near the maximum
were a little erratic, probably owing to the fact that at
that part green was distinguished, the rest of the spec-
trum being grey or brownish grey to N. W., and they
therefore presented no difficulty in comparison with
the white beam. Using rays on each side of the maximum

UNCOMMON COLOUR BLINDNESS 345

to form the equations, the factors of reduction of the
curve to compare it with the normal curve are obtained.
Taking SSN.'s 56 and 40, we form the first pair of

equations from the luminosities m Table LIY. and that of
N. W.'s luminosities. As before, the right hand of theequa-
tions are formed from the RS. numbers in Table LIV. —
50 ~ 302 = 477y 50 - 6^-5 - 25-8y

This ^ves —

y = 073 a = 0-51
Another pair of equations can be formed from SSN.'s
54 and 44 —

80 - 522 = 72-4y 75 - 882 = 31-3y

which give —

y-0-69 2=0-56

346 RESEARCHES IN COLOUR VISION

From SSN/s 52 and 42 we get—

96-70z = 80-6y 62-5- 722; = 34-6y

which make—

y = 074 2 = 0-52

From SSN. 8 58 and 46 we get—

21-102; = 20-8y 87-90z = 5A'2y

which make —

y = 0'77 z = 0'5

From SSN/s 60 and 38 we get—

7-32 = 7y 36-462; = 17-5y

which make —

y = 0-78 2 = 0-57

The mean of the different values of y is . .0-74

And that of the different values of 2 is . . 0*53

For the sake of simplicity, we may take the values
as 3/ = 0*75, that is, x (the red sensation) is 0'25 of the
normal, 2; = 0*50. In Table LXI. these values are em-
ployed. Column V. gives the theoretical curve derived
from Table LIV. containing the colour equations.

Comparing columns IV. and V. together, we see that
at the position of maximum luminosity the theoretical
values differ from those obtained from the readings,*
the mean of which was taken. A further examination
of these two columns also shows that at the violet end
of the spectrum the luminosity values obtained by N. W.
are much larger than given in the normal curves.

The luminosity of the blue sensation is very small

^ Had the evidently low readings been omitted when calculating the
mean luminosity value, the two would have taUied well.

UNCOMMON COLOUR BLINDNESS 347

compared with the luminosities of the red and green,
and is negligible as far (say) as SSN. 40, but from SSN. 25
to the violet end of the spectrum the luminosity of the
blue sensation plays a larger and larger part in the total
luminosity of each scale number.

We have already found the factor of the red sensa-
tion (which we see from the table forms part of the
normal violet). If, then, from the luminosity values
obtained in this region of the spectrum by N. W., we
subtract her red sensation, and also her green sensation,
the residue will be due to the blue sensation, which can
be compared with that existing in the writer's vision
within the yellow spot.

Taking her readings from SSN.'s 25 to 0, we obtain
the following result : —

Table LXII.

SSN.

N. W.'8
Luminosity.

•1

N. W.'a

lioduced Red

Sensation.

•014

GS.

• • •

N.W.'sBS.

Nonnal
BS.

N.W.'s BS.
Normal BS.

4

•086

•022

10

•75

-m

• • •

•64

•1

6-4

16

1-25

•11

• ■ •

114

•16

7

20

1-5

•2

•1

1-26

•234

5^8

25

2-5

•32

•84

1-34

•25

6^4

If we lay down the luminosities shown in a curve,
and draw a freehand curve between the points, we get
= 0-15, 10 = 07, 15--= 1-1, 20 = 1-65, 25 = 2-6 as ordinates,
and the resulting ordinates of N. W.'s blue sensations
are six times larger than those of the normal curve.
This gives a very good clue^ to her naming the colours of
the spectrum as given .

^With a normal eye fatigued by red to produce *25 RS., particularly
with an excess of blue sensation, the colours seen would not be far different
from those of N. W. See Chapter XXV.

348 RESEARCHES IN COLOUR VISION

An endeavour was made to form a series of colour
equations with her eyesight by placing three slits in
different parts of the spectrum, but without success,
although a match with white was made in two positions. \
One slit was placed in the orange red at about 52 (X 6000) |
of the scale, another at E, and the third at G, and white i
light was formed, though her match was so erratic that
it was useless to measure the apertures. When the slit I
in the violet was covered up, a white patch being along-
side as a comparison, she called the mixture of red and
green "brownish green"; when the slit in the red v^as
covered she called the mixed light of green and violet
" green " ; and when the green slit was covered up she
called the purple colour a ** different kind of brown."

When the first slit was moved into the red near the
lithium line she called the colours "green," whenever the
green slit was uncovered. A piece of signal-red glass
of the London, Brighton, and South Coast Railway M^as
placed in the white reflected beam, forming a red patch,
and a patch of the blue scale at No. 30*5 (X 4862) was
placed alongside, and she matched them in luminosity '
and in colour. (The dominant colour of the signal glass
in question was X 6220.) She finally was tested with
colour discs : *—

One being red with dominant wave-length (R) X 6150
Another, emerald-green (G) „ ,, . X 5373

And the third, French ultramarine (U) ,, . X 4700

To make white she required —

130 G + 113R + 117U = 72 W + 288 B. (Black cor-
rected for white light reflected.)

' The colours of the discs were all impure colours, and each colour stimu-
lated all three sensations more or less. A reference to Chapter XVI. will
show how it was that the discs matches could be made. Chapter XXIII.
should also be studied in connection with them.

UNCOMMON COLOUR BLINDNESS 349

She was then tried with the blue and green discs
alone, and made a match —

258 U + 102 G = 65 W + 295 B (corrected).

An attempt was made to match with the green and
red discs alone, but this failed.

She matched the red disc alone with black and white,
and also the blue disc alone —

360 R = 56 W + 304 B (corrected),

360 U = 60 W + 300 B (corrected).

With any proportion of R and U mixed together she
matched a grey of approximately the same intensity as
above, as it might be supposed she would from the last
two equations.

A Case of Coloiir Blindness with Great Excess of

Pigment}

The next case (M.) is one in which the amount of
pigment in the retina is so great that it practically cut
off a large portion of the blue end of the spectrum.
M. was examined more than twenty years ago by the
author, when the sensation curves had not been worked
out. It was believed at the time to be a case of violet
blindness. His description of spectrum colours was most
remarkable. He only saw two colours, red and black.
He called all green and blue, black ; green, however, he
called bright black, blue being darker black ; yellow he
called white. At SSN. 52 on the scale he saw a "little
red," at SSN. 50 " no colour " ; his neutral point where
he saw the spectrum colourless would be about 49*5, or
about X 5800.

His luminosity curve is given in the following table : —

* Paper No. 4.

350 RESEABCHES IN COLOUR VISION
Tiinj LXIII.

8c*I« Number.

UNCOMMON COLOUK BLINDNESS 351

Taking SSN.'s 52 and 46, we get, according to the
first method in the last chapter, —

96-52^=15-36?/

87-4l2 = 32-7y

y=-54 ; that is, M. has '46 GS.

Taking SSN/s 56 and 44—

50-27^ = 2-25y

75- 322 = 30-8

y = -56, orM. has "44 GS.

Finally, taking SSN/s 54 and 42—

80-42-52 = 7-62/

62-5 -232 = 27-751/
2/= -62, or M. has '38 GS.

On considering the whole curve, and where the
absorption due to pigment may commence, we shall not
be far wrong if we make M. to have -46 to -5 GS. The
curves of a '46 GS. would lie closely on that of M. on
the red side of the maximum and the absorption would
begin to show at SSN. 46. The extinction curve of M.,
which was rather erratic, shows that that part of the
curve in the red resembles that of a normal eye, but in
the parts where the absorption is powerful the normal
eye shows the fundamental light as at least 180 times
greater than M.'s.

Case of White Sensation only to one Eye, ivhilst Normal

to the other Eye.^

Miss W. was brought to the laboratory by Dr.
Lindsay Johnson. The right eye was apparently normal
for colour, but with the other she saw nothing but
shades of white.

» Paper No. 17.

352 RESEARCHES IN COLOUR VISION

Miss W., it appears, has had a slight stroke of
paralysis, which affected her left side, and subsequently
she discovered that colour sensation in the left eye had
disappeared. The cause, from an examination by a
specialist, seemed to be atrophy of the optic nerve.

She was examined with the spectrum colours, and
her left eye found to be totally blind to every colour,
though her perception of light was fair. She had very
little difficulty in comparing the luminosity of the most
brilliant spectrum colours with the white patch of light
placed alongside them. In making the measurements
she experienced a certain amount of fatigue, but, by
resting the eye for short intervals, her readings were
very constant. The following is the table of her read-
ings : —

Table LXIV.

Scale No.

Wave-
length.

Readings.

63

7082

62

6957

1

60

6728

7

58

6520

18

57

6423

28

56

6330

43

54

6152

76

62

5996

90

50

5850

95

48

5720

93

46

5596

83

44

5481

71

42

5321

58

40

5270

46

38

5172

32

36

5085

21

34

5002

12-5

32

4924

7

30

4848

4-5

28

4776

3

25

4675

1-5

20

4518

0-4

19

4488

Remarks by Mias W.

Both colour and white patch appeared
as white throughout the spectrum.

UNCOMMON COLOUR BLINDNESS 353

At 19 the light perception was so diminished that
she could not match the grey. Her light perception
extended further into the violet (as white) beyond this
point, as the subsequent measures of her extinction of
light showed conclusively.

The orange sodium light of the spectrum was thrown
on the apparatus therein described, of a luminosity of an
amyl lamp 1 foot oiF, and the slit giving this brightness
remained unchanged throughout the examination, and
was moved through the spectrum till a position was
reached where all light was just extinguished. Her
perception of the point of extinction was very acute.
Rotating sectors were placed in front of the apparatus,
set at different angles, so that the amount of reduction
of the luminosity of the spectrum was known at once.

Table LXV. — Scale Readings of Light Extinction.

lAghi coming^ through
the slit reduced to-

Slit moved towards
the violet.

Slit moved towards
the red.

No reduction.

15

63-7

^ intensity.

20-7

62-4

} »,

21-7

60-9

} .

23-2

48-7

A n

26-7

46-7

^6 »

34-7

44-2

1^ .

—

40

The extinction of light with the full aperture to the
writer was with the size of the patch employed at 57*9.
At 57 '9 the luminosity of the spectrum is 022 that at
the D line, and as the light on the screen at the end

z

354 RESEARCHES IN COLOUR VISION

of aperture was 1/620 that falling on the instrument
originally, it follows that the extinction to a normal
eye when the light of 579 (X 6510) was 022 620 or
0'000355 of an amyl lamp placed at 1 foot from the
screen.

At D, if 71/100,000 of an amyl lamp illuminated a
screen 1 foot off, it is invisible to Miss W. With normal
vision, if the screen be illuminated with 7/100,000 of the
same light, it just becomes invisible. She has there-
fore about 1 10 of the light of normal vision in the
colour blind eye.

A Probable Case of Monochromatic Vision.

About July 1892, a case of colour blindness quite
unusual was examined by the author and published by
the Royal Society. ^ B. C, as he was called in the paper,
was then a youth of nineteen, who had been serving at
sea. His form vision was perfect, and he was not night
blind, though he stated that on a cloudy day his vision
seemed to be more acute than in sunshine. There is
reason to believe that some of his ancestors were colour
deficient. Being tested with the wool test (see Chapter
XXVI.), he matched all colours with one another. He
called the lighter colours '* dirty " or " pale " blue, terms
which were found to be synonymous. In the spectrum
he called every colour blue except the bright yellow,
which he called white ; but when the luminosity of the
colour was reduced, he pronounced it a good blue. So
with white, he called it white when bright, but as its
luminosity was reduced it passed through the stage of
'^dirty blue" to full blue.

* Paper No. 4,

UNCOMMON COLOUK BLINDNESS 355

Testing him -with colour discs (see Chapter XI.) he
made the following matches : —

360 red matched in bright-jg^g ^ ^^ ^^ ^j^j^
ness and colour .

360 green matched in bright-
ness and colour .

258 black + 102 white.

360 blue matched in bright-lg^j^ y^^j^ ^^ ^j^j^^
ness and colour .

With these proportions he emphatically stated that all
were good blues, and that the inner disc and outer ring
were identical in brightness and colour.

The luminosities of the red and green to the normal
eye are not very different. The equation of red and
green shows that the green is much brighter to B. C.
than the red or the blue. Further, it must be remem-
bered that the red contains a considerable percentage of
green sensation and the blue a large quantity of green
sensation and also some red ; whilst the green also has
more than half of red sensation (see Chapter XVI.). As
B. C. only saw one colour, it seemed likely that the one
sensation he felt was the green. The white, of course,
would appear green, and is quite independent of the
name he gave it. His curve of luminosity is a remark-
able one, and is given in Table LXVI., and is shown in
the figure. A great falling off in the luminosity when
compared with that measured by the normal eye will be
noticed both in the red and the blue of the spectrum,
and, as before said, it seemed probable that his chief
colour sensation was the green.

The luminosity of a spectrum to his eyes was measured
with some diflficulty at first, but afterwards, when the
terms " dirty " or ** pale blue " applied to a colour had been

356 RESEARCHES IN COLOUR VISION

explained by the disc experiments, the measures were
made with some degree of accuracy. The method

adopted was to diminish the white light illuminatine one
shadow in the colour patch (see Chapter VIII.) to a point

UNCOMMON COLOUK BLINDNESS 357

which made it appear a good ** blue " to him, when a
slight alteration in the intensity was sufficient to secure
equality of luminosity between the white and the colour.

Table LXVI. — B, C.^s Luminosity and Extinction,

I.

II.

III.

IV.

V.

VI.

Comparative

SSN.

X

Adopted

Readings of

Extinction in

A-

Persistency
Curve.

Luminosity.

Luminosity of

Eltinction
when each Ray

Maximum 100.

Extinguished
is of Equal

100000

60

6728

Luminosity.

5500

2-3

•5

27-5

58

6520

2800

4-5

2

56

56

6330

1500

8-3

6

90

54

6152

950

131

11-5

109-2

52

5996

580

21-6

21-5

125

50

5850

350

36

37

129-5

48

5720

215

58

60

129

46

5596

140

89*3

92

129

44

5481

126

100

100

125

42

5373

150

83

85

127-5

40

5270

215

59

45

96-7

38

5172

290

43

21-5

72-3

36

5055

380

33

11-5

43-7

34

5002

500

25

7

35

32

4994

650

19

4

26

30

1

4848

850

14

2-5

23-3

1 28

4776

1100

11-4

2

22

! 26

4707

1500

8-3

1-5

22

; 24

4639

2000

6-2

1

20

22

4578

2700

4-6

•5

13-5

20

4750

4750

■ . •

• • ■

• • •

Treating the curve of B. C. as we treated that of M.
and N. W., we find that it answers very fairly to the
green sensation curve. Any small divergence from it is
probably due to the errors of observation by an untrained

358 RESEARCHES IX COLOUR VISION

obeerver. The highest factor of deficiency obtained is
1 '2 and the lowest '95 for the red sensation. The first
is of course impossible, but a mean of all the factors
thus obtained is closely unity, which shows that he
possesses no red sensation.

A further test of his colour sensation was made by
taking the extinction of the various rays of the spectrum.
His observations were fair except on the violet side of
the F line where they became slightly erratic, but by
requesting him to use the whole retina to obtain the
last glimpse of light, a very concordant curve resulted.

In the figure the extinction of the various places in
the spectrum are indicated by x , and the extinction has
been taken from the freehand curve drawn as nearly as
possible through the several points.

When the " persistency curve " was made, it agreed
closely in the green and yellow with the luminosity
curve, which stopped when not far in the blue.^ As the
whole retina was employed in the extinction observations,
it indicates that the falling off of luminosity in the blue
part of the curve is not due to excessive pigmentation in
the yellow spot, and seems to point to an absence, total or
nearly total, of the sensation of light (in contradistinction
to colour). If we turn to Table XIIL, col. IV., page 150,
we shall find that at SSN. 44 the extinction of colour to a
normal eye takes place when it is '0027 of the luminosity
of the ray when D is 1 candle at 1 foot distant from the
screen. In Table LXVI. it is shown that B. 0. loses all
sensation of light when the same ray under the same
conditions is reduced to '00125. This last is of the
same ** order " as the first. If this is a case of mono-
chromatic vision, it is quite a different kind to that
recorded for P. and Q., since their place of maximum

^ The deviations of the persistency curve is shown by the dotted line.

UNCOMMON COLOUR BLINDNESS 359

luminosity diflTers largely from that of B. C. It appears
that whilst the former only have the sensation of light
and not that of colour, B. C. appears to have the sensa-
tion of green and probably the absence of the sensation
of light.

CHAPTER XXV

ON COLOUR FATIGUE

We will next consider if the results of fatigue of the
retina by different colours bears out or disproves the
trichromatic theory.

After Images.

In its simplest and qualitative form fatigue is
shown when an eye steadily gazes at a spot of colour on
a black ground. When the eye is then turned to a grey
paper, an image of the spot will show itself, and travel
over the paper as the eye moves, and will be of a
colour complementary to the real colour of the spot.
When it is a red spot that is looked at, the after image
of the spot on the grey paper will be a bluish green,
though pale. If it be an emerald green spot, the image
on the grey paper will be pink, and so on. The explana-
tion is perfectly simple. When the red spot has been
steadily looked at, its image falls on a small portion of
the retina and principally acts on the red receiving
apparatus. If colour vision be connected with the
chemical decomposition of some red, green, and blue
sensitive substance, then the prolonged gaze decom-
poses a certain quantity, mostly of the red. and the
sensitiveness to red becomes less and less. When the
eye is turned to the grey paper (degraded white), the
light from this ordinarily would stimulate each of the

360

ON COLOUK FATIGUE 361

three receiving apparatus equally. The red apparatus
not having recovered the full sensitiveness in the spot
on the retina on which the red image fell, it follows that
the green and the blue are stimulated much more than
the red. As the red sensation only acts partially on the
red fatigue spot on the retina, the after image seen on
the grey paper is a pale blue-green image. For similar
reasons, when other coloured spots are made to fatigue a
a spot in the retina, we have the complementary colour
when looking at the grey.

With the colour patch apparatus we can study
fatigue qualitatively and quantitatively in a very simple
manner. We can place a patch of white light on a
small square surface, and fatigue the whole of the
retina by looking at the surface of the prism through
one or other of the slits, placed in the various colours.
Closing one eye and looking at the red ray with the
other, fatigue is induced in the latter. On looking at
the white patch, it will be found to take the comple-
mentary colour to the red, viz. bluish green ; the brighter
the light seen on the surface of the prism, the less pale
the blue-green will appear. Using the unfatigued eye,
it will be seen that the white is unchanged.

By using a second square on which to receive light,
a patch of blue-green light mixed with a varying amount
of white can be arranged. By use of a kind of stereo-
scopic arrangement, which we describe later, the white
patch can be viewed by the fatigued eye, and the colour
by the unfatigued eye, and after a few trials the two can
be made to match. Instead of the white on one patch,
we can place, say, a greenish yellow near the D line
about SSN. 487. At this point the two sensations of
red and green, according to the writer's measures, are
equally stimulated — that is to say, it is the place wher6

362 RESEARCHES IN COLOUR VISION

the red and green sensation curves cut when made of
equal areas. If the trichromatic theory holds good, an
eye fatigued hy the red should show this colour as no
longer yellowish, but decidedly greener ; and if fatigued
by the green, decidedly redder. When this experiment
is carried out, the results are those to be anticipated.
By throwing a greenish patch upon the second square
from a second spectrum, the green can be matched, as
is shown later on. In the same way other colours can
be examined, and in all cases the confirmation of the
trichromatic theory seems to be complete.

The reverse operation can be carried out : the fatigue
may be made by the colour which is to be examined.
Take the ray at SSN. 48 '7 and fatigue one eye with it,
and then see the effect it has upon a pure red patch.
It will be found that the red has lost considerable lumi-
nosity, which can be. verified by observing immediately
afterwards the patch with the unfatigued eye. It has
been suggested that there is a sympathetic action
between the two eyes, but these experiments and
others leave no doubt that the sympathetic action is
not recognisable. Using the stereoscopic arrangement
described later, the luminosity of a red patch placed
next that observed by the fatigued eye, can be matched by
it in luminosity. When the patch caused by the green
ray at (say) SSN. 37*5 is observed with the fatigued
eye, it will be foimd to be of diminished luminosity and
to have a slightly more bluish tint than it is to the un-
fatigued eye, which is again what would be predicted
from the trichromatic theory. The fatigue may also be
caused by a prolonged gaze at a patch of colour. The
general results that are obtained from these fatigue ex-
periments in the spectrum are as follows. When the eye
is fatigued by red, the red itself is reduced in lumi-

ON COLOUR FATIGUE 363

nosity ; the orange becomes yellow, the yellow greener ;
whilst the green, owing to the inherent white, becomes
a bluer green ; the blue-green is not so much affected ;
the blue becomes greener and the violet becomes bluer.

When the eye is fatigued by green, the red remains
unaltered; the orange becomes redder, as does the yellow ;
the green becomes paler, and at one part nearly white ;
the blue-green becomes bluer, the blue more violet, and
the violet unchanged.

Fatigue by a patch of blue is more difficult to in-
duce. The principal change is in the blue-green, which
becomes greener, and the violet redder. As the blue ray
which answers best to the blue sensation is mixed with
some 80 per cent, of white, and is only feebly luminous,
it is not hard to understand the feeble nature of the
fatigue which is induced. In reference to the fatigue
produced by the white, it is only necessary to advert to
an experiment with a white wafer on a ground of black
velvet. When steadily gazed at by the eye, which is
then turned to a grey surface, it will be found that the
image of the white spot will appear darker than the grey.

Fatigue hy Extremely Bright Colours.

Coming next to fatigue by more intense colours, we
must refer the reader to the most suggestive paper by
Dr. Burch ^ on ** Artificial Temporary Colour Blindness."
By fatiguing the retina with extremely bright colours,
complete temporary colour blindness was apparently in-
duced. In order to get red fatigue, he employed a
burning-glass of 2 inch focus, placed at such a distance
from- the eye that the pupil was filled with direct rays
of the sun after passing through ruby glass backed with

* PhiX. Trans., Series B., vol. cxci. (1899), pp. 1-34.

364 RESEARCHES IN COLOUR VISION

a magenta-stained film. For green fatigue he employed
three thicknesses of green glass coloured with cupric
oxide; and a tank filled with an ammonio-sulphate of
copper served to give violet fatigue (Dr. Burch came to
the conclusion that his vision required a fourth sensation
of violet).* For blue fatigue he reverted to the blue of
the prismatic spectrum. By fatiguing the retina for a
sufficient time with these different colours he became com-
pletely red, green, blue, or violet blind, and describes what
effect such colour blindness had on the colour of objects.
So far as the red and green sensations are concerned, it
appears that the effect produced is that experienced by
the eye which is congenitally totally blind to one or
other of these sensations. In the next pages it is not
proposed to show the result of extreme fatigue of the
retina, but only of fatigue sufficient to show that incom-
plete colour blindness can be imitated by it in an eye of
normal vision.

Use of a Stereoscope.

For use in the above and subsequent investigations
of the effect of retinal fatigue by the different colours of
the spectrum, a special arrangement had to be made for
viewing the white or the colour with the fatigued eye,
and at the same time any other colour with the un-
fatigued eye. It was also convenient that the colour
which was to be the fatiguing colour should be brought
to the place which the eye would occupy when viewing
the coloured or white patch.

Fig. 97 will show the arrangement. A and B are slits
in the double spectrum apparatus (see p. 44) with the
collecting lenses L and L' in situ to throw patches of

^ '^ The sensation which in this book is called the blue sensation seems
to be intermediate between Dr. Burch's blue and violet."

ON COLOUR FATIGUE 365

any desired colours on the different white screens C and
D. (If a white on the screen D is required, the slit B
is removed and the whole spectrum is collected by L'.
The luminosity of the white can be reduced by sectors
or other means.) K is a thin screen pointing as shown,
so that its direction, if produced, would divide the
distance between C and D equally. E' and E^' mark
the position the eyes occupy. For the sake of comfort,
the forehead rests on a pad fixed to K. With this

r /H

B

Fig. 97.

arrangement the right eye sees the patch D only, and
the left eye the patch C only. This arrangement is
made for use when the right eye is to be fatigued.

To bring the fatiguing ray to the eye, a third spec-
trum with a movable slit is formed at H by utilising
the reflected beam of the colour patch apparatus (which
is ordinarily employed as a comparison white light) to
form a spectrum with a second set of prisms. The ray
from this spectrum, coming through a slit, passes through
a lens l/\ and falls on a mirror M.^ The mirror M is

* The use of the mirror was suggested by Dr. Watson, to allow the head
and eyes to remain in one position during the whole of one observation.

366 RESEARCHES IN COLOUR VISION

pivotted at P, and can be brought against K, leaving an
uninterrupted view by the right eye of D. The lens \J"
is inserted in the beam so that the whole of the pupil
may be fatigued.

Should it be desired to have a less intense light for
the fatigue, a white surfisu^ can be substituted for the
silvered mirror, a colour patch being formed on it. To
use the apparatus, the eyes are placed in position, and
the distance apart of the C and D is so arranged that
the patches of colour appear to touch. When it is
desired that the same colour shall be in both patches,
the Scale Nos. of A and B being known, this can be
easily done and a confirmation of the correctness be
made by making C and D overlap and placing a rod in
the path of the beams, using unfatigued eyes to form a
judgment. The slits in each can be so adjusted that the
two patches are equally bright.

Qualitative Ohsei^ations.

The first experiments carried out were qualitative,
with the fatigue in the right eye induced by the colour
patch at M. Patches equally bright were thrown on
C and D with unfatigued eyes. The two eyes were
then placed in position, and the right eye fatigued by
gazing at the colour patch on M for 30 seconds. A
twist given to P uncovered D and the effect on the
colour noted. The following are the notes of the
qualitative observations, both patches remaining
unaltered.

The luminosity of the ** fatiguing " patch was about
2 candles 1 foot from the screen. The changes noted
are when the fatigued eye is compared with the un-
fatigued.

ON COLOUE FATIGUE

367

Table LXVII.

SSN.

69-8

67-6
50-6

42-8

37-6

81-2
16-6

All the
yiolet

Red Fatigue.

Fatigue with SSN. 50*6
(D).

Fatigue with SSN. 42*8.

Same colour, but darker

I i II li 11
Greener and slightly

darker
Oreen ; slightly bluer

M II }f

No perceptible change
Bluer than unfatigued

Much bluer and darker

No change; a little

darker
Colour a little darker
No change, only darker

Bluer and darker

Much bluer and rather

darker
Slightly bluer
No apparent change in

colour, but darker
Bluer and darker

A little darker

Slightly darker

Darker ; no change in

colour
Slightly bluer

Bluer and darker

No visible change in

colour
Slightly dimmer

Percentage Composition of Spectrum Colours in tei^ms of
Equal Stimulus of Sensations to form White.

Any observations made to secure quantitative results
with a fatigued retina will be best discussed with the aid
of a percentage table of equal stimulation of the three
sensations to form white. This will be found in Table
LXVIII., for a spectrum of the arc light with horizontal
carbons with which all the observations in this chapter
were made as a source.^

Fatigue hy White and the Law of Fatigue.

In regard to the above table, and the fatigue which
will be referred to it, attention must be directed to the
fact that if the fatigue of the retina is by a white beam,
a similar white observed by such an eye will be merely
darker than the latter, and no change in colour will be

^ In this chapter the ordinates of the sensations for equal stimulus to
give white are denoted by R'S., G'S., and B'S., instead of RS., GS., and
BS., which are the symbols for the '' luminosity '' sensation.

368 RESEARCHES IN COLOUR VISION

observed. If the two whites are compared together by
the two eyes, the fatigue white will appear a dark grey.
It follows, then, that the three sensations on the " equal

Table LXVIII. — Calculated from Table XL,, rolumns 4, 7, awl 8,

page 244.

Ratio of

ssx.

X

R'S.

G'S.

B'S.

R'S. to G'S.

•

1

_ _ _ .
7217

100

(OS. = 1).

64

• « •

• • •

100

1 62

6957 .

100

• • •

• • •

100

60

6728

100

• « ■

• • •

100

58

6521

97

3

■ • •

323

56

6330

90-2

9-8

...

9-21

54

6162

81

19

• ■ •

4-21

52

5996

70-7

29-3

•••

2-41

50

5850

57-5

42-5

■ ■ •

1-353

48

6720

47-2

51-2

1-6

•922

46

5596

41-6

56

2-4

•743

44

6481

37-5

59*5

31

•631

42

6373

33-9

61-2

4-9

'554

40

6270

29-7

621

8-2

•486 :

38

5172

25-8

60

14-2

■43

36

5085

21-8

56

22-2

•369

34

5002

16-8

47-1

361

•351

32

4924

11-6

34-2

54^2

•339

30

4848

7-8

22

70-2

•355

28

4776

5-7

14-1

80-2

•404

26

4707

4-3

7-7

88

•558

24

4639

3-45

3-67

93

•94

22

4578

3

1-66

95-4

1-81 ;

20

4617

2-71

•68

96-6

384 1

18

4469

2-47

•35

97-2

706

16

4404

2-32

•16

97-6

16-5

14

4349

2-22

• • •

97-8

...

1

stimulus " scale, when fatigued by white, will cause the
three ordinates to be equally diminished. Again, if the
fatiguing colour be that at SSN. 48 6, where the R'S. and
G'S. are equal, then the R'S. and G'S. in any colour in

ON COLOUB FATIGUE

370 RESEAKCHES IN COLOUR VISION

which there are both sensations will be equally affected.
Thus, if the fatigue induced makes the R'S. and G'S., one-
half the unfatigued sensations, and a colour, of which the
normal composition is, say, 1 R'S. to 2 G'S., is to be
observed. This fatigue will only make the colour ^ R'S.
to 1 G'S., or the proportion of R'S. to G'S. remains the
same. But if the eye fatigued with this colour observes
a red in which there is no G'S., the only effective
fatigue will be that of the R'S.

Examples of Fatigue.

Between SSN/s. 40 and 42 there is a ray in which
the ordinates of the G'S. is twice as large as that of the
R'S. If this be the fatiguing colour, any other colour
observed with the fatigued eye (except in the extreme
red) will be altered in hue as the fatigued R'S. and
G'S. are unequal. As an example, let the fatigue be
such as to reduce the original R'S. to \ R'S. : the
fatigue of the G'S. will be twice as great as that of
the R'S.

Let the patch of colour be, say, that of SSN. 50,
whose composition is R'S. 545 to G'S. 42*5. The fatigue

will make the R'S.=-^ or 13-6, and the G'S. = ^

4 2

or 21-25. This is a ratio of G'S. to R'S. of 1 to "64 (see
column V. of Tafcle LXVIIL), or the colour the fatigued
eye would see would be about SSN. 44 — that is, the
yellow of SSN. 50 would be seen by the fatigued eye
as the green of SSN. 44.

At a SSN. between 38 and 36 (see Table LXVIIL)
there is a colour in which the ordinates of the R'S.
and the B'S. are equal, but the G'S. has an ordinate
which is largely in excess of the other two. Fatigue

ON COLOUR FATIGUE 371

given with this colour is practically fatigue with the
amount of G'S. which is in excess of the amount
necessary with the other two sensations to form
white, except where there is no R'S. or B'S. in the
colour.

An actual measure of a match made will be of use
to illustrate the observations recorded in the next
table.

Fatigue was made by a fairly bright red at SSN.
59*8, which is pure red sensation, and the colour to be
matched when observed with the eye thus fatigued was
SSN. 48-7 where the R'S. and G'S. are equal.

The match made with an eye unfatigued was found
to be at SSN. 34'3, but it was very pale. The R'S. had
therefore diminished from 1 to '35 nearly. Previously a
measure of the luminosity of SSN. 59*8 had been made
with the fatigued eye and the unfatigued, and found to
be as 10 to 29 or as '34 to 1. Now, there is practically
no white in the colour of SSN. 487, but a large amount
at SSN. 3 4 •3. When the fatigue changes the former
into a green, the nearest spectrum colour to match seems
very pale, hence the green produced by fatiguing the
eye is a green much purer than any spectrum green
seen with an unfatigued eye.

Matching the Spectrum Colours ivhen Fatigved

hy Red.

The following is a good illustration of the matches
that can be made when the eye is steadily fatigued
with a pure red of constant brightness, and the
pupil is submitted to its action till a fair balance of
fatigue and recovery is struck before an observation is
made.

372 RESEAKCHES IN COLOUR VISION

Table IjXTX.— Fatigue by the Red Ray at SSN. 59-8.

Fatisraed Eye Observed >

Unf Atisrned Eje Match >

8SX.

SSN.

68-6 •

Unchanged

56

53*4

63*34

49*64

60-7

39

48-6

34-6

46-4

32*2

42-8

33*5

401

34

37-6

32

32-2

Unchanged

29-6

31-2

27

3015

21-7

29*6

16 5

20

The first point to call attention to in the above is
that from SSN.'s 56 (in which there is only a small
quantity of G'S.) to 32 '2 (where there is no change
in hue capable of being measured accurately), the
matches are throughout lower in SSN/s than the fatigue
colour. In column VI. , Table LXVIII., it will be seen that
at that SSN. (32*2) the ratio of red to green is at its
minimum. From this number to SSN. 16 the readings of
the matches are always higher. In both these divisions
of the spectrum the smaller luminosity of the red in the
fatigue colour is therefore always shown in its match.
[This is a direct general confirmation of the truth of the
percentage curves of the three sensations, and therefore
of the luminosity curves from which they were derived.]

* The numbers in this scale are apparently awkward places in the
standard scale, to which everything is referred ; so far they are the num-
bers which are derived from a temporary scale. The fatigue colours were
whole numbers.

' The wave-lengths of SSN.'s of whole numbers will be found in Table
XXXVIII. and other tables.

ON COLOUK PATIGUE 373

Percentage Composition applied to Colour Blindness.

[Attention must be called to the fact that the "equal
area" or "equal stimulation" curves apply not only to
normal vision, but also to colour blindness. The differ-
ence between the two is that the ordinates of either one
or two of the colour blind's luminosity sensation curves
have to be multiplied by a higher factor or factors than
is applied to the normal vision curves. It is due to this
that the following method of finding the factor of fatigue
is possible. When the factor of any sensation fatigue is
found, the equal area curve for that sensation can at
once be calculated.]

Obtaining the Factor of Fatigue.

Studying each observation in detail, so far as is
necessary, it will be found that the amount (or factor)
of fatigue of the retina is a fixed one when the observa-
tions are made as described. To exemplify the method
of calculation, one of the observations may be taken in
which there is a large proportion of R'S. to G'S., say
SSN. 53-34. The match to this colour is at SSN. 49*64.

SSN.'s 53-32 and 49*64 have for their sensation com-
positions —

R'S. Go.

SSN. 53-32 78 + 22
SSN. 49-64 56 + 44

The only alteration made in SSN. 53-32, when the
fatigued eye observes it, is a diminution of the R'S. ; the
G'S. remains unaltered. In the match there is, of course,
the normal proportion of R'S. to G'S. If, then, we make
the G'S. of the fatigue and the match composition the

374 EESEARCHES IN COLOUR VISION

same, we can directly compare the R'S. in the fatigue

colour with its R'S. when the eye is not fatigued.

In the case in point it happens that the G'S. in the

match colour is exactly half of that in the fatigue

colour. Its composition is equally well expressed as

RU28+G'S. 22.

28
The fatigue R'S. is therefore ^q ^^ *^® normal R'S.

in SSN. 53-32, and the factor of fatigue is '36 RS.

Taking SSN. 507, its match is at SSN. 39. The
composition of these two are —

and

or

R'S. G'S.

(SSN. 507) 61-5 + 38-5

R'S. G'S. B'S.

(SSN. 39) 28 + 59 + 9-5

R'S. G'S. B'S.

18-3 + 38-5 + 6

From this we see that the factor is '3 of R'S.
In the same way, taking SSN.'s 486 and 34*5 having
composition of —

R'S. G'S. B'S. R'S. G'S. B'S.

49 + 49 + 2 and 17-5 + 60 + 32

we obtain a factor of 3 R'S.

At 45*4 we have a match with 32*2, with composi-
tion of —

R'S. G'S. B'S. R'S. G'S. B'S.

40 + 57 + 2-5 and 12 + 34-5 + 53

The factor for this, after deducting the white present
in the fatigue colour, gives a factor of -37.

ON COLOUK FATIGUE 375

At SSN. 56, when the match is at SSN. 53-4, the
compositions are —

R'S. G'S, R'S. G'S.

90-2 + 9-8 and 77 + 23
From these we get a factor for R'S. of 36.

Match Colours from SSN. 37*5.

If we examine the match to SSN. -42*8, which is at
SSN. 33*5, we have the following compositions —

R'S. G'S. B'S. R'S. G'S. B'S.

35 + 61+4 and 16 + 44 + 40

After deducting the white in fatigue colour from both,
we have —

R'S. G'S. White. R'S. G'S. B'S. White.

31 + 57 + 4 and 12 + 40 + 36 + 4

Treating these compositions in the same way, we get
as a factor '55 R'S. This increase requires an explana-
tion. The G'S. to R'S. in the match colour is 1 to '3, and
turning to Table XL. it will be seen that there is no colour
which has so low a ratio ; hence the eye has to do the
best it can in finding a match. The same is the case
with the next two numbers. In Chapter XVII. it will
be seen that from SSN.'s 36 to about 56 the blue which
is added by the white light is ignored by the eye, and
the hue of a colour is judged by R'S. and G'S. only, but
that after SSN. 34 is passed (towards the blue) this no
longer holds good. It is the same also in the matching
of the ** fatigue colours." For instance, let us consider
SSN. 32*2, which matches itself. Its composition is —

R'S. G'S. B'S.

12 + 34-5 + 53
The average factor for the R'S. is '34.

376 RESEARCHES IN COLOUR VISION

If we multiply the 12 R'S. by this number, we have
R'S. 4*1 in the fatigue colour.

There is therefore no change in it except a greater
degree of paleness in —

RU GU BU

12 + 34-5 + 53

and

R'S. G'S. B'a

4-1 + 34-5 + 53

Converting these two equations into sensation and
white, we get—

G'S. B'S. W.

22-5 + 41+36

and

GU BU w.

30-4 + 49 + 12-3

The mixture of white and the green and blue sensa-
tions are paler in the first than in the second equation,
but the general hue would be the same. This is the
case in all degrees of red fatigue tried ; the match at
SSN. 32-2 is invariably that SSN. itself, the only differ-
ence being paleness, the BS. differing but little in each
equation .

One more matoh must be considered, viz. that at
SSN. 21*7, which when it was a fatigue colour was
matched at SSN. 29-6.

The composition of SSN. 21-7 is —

R'S. G'S. B'S.

2-9 + 1-6 + 95-5

the R^S. hmig in excess of the G^S.

Taking '34 as the factor, we get R'S. 1 (nearly), and
this is all used up in making the white, and the com-
position becomes —

W. G'S. B'S.

3 + -6 + 94-5

ON COLOUR FATIGUE 377

SSN. 29 '6 has a composition of —

R'S. G'S. B'S. White G'S. B'S.

6-6 + 15 + 78-4 or 19-8 + 8-4 + 71-8

the G'S. being greater than the R'S.

Reducing the last composition to make the G'S. the
same as that of the fatigue colour, we get —

w. G'S. B'S.

1-4 + -6 + 5-5

The ratio of white to G'S. is not very different in the
two, and the eye ignores the excess of B'S. in the first.
[It must be recollected that the luminosity of the B'S. is
110 times smaller than that shown as B'S.]

In this case the fatigue alters the ratio of R'S. to
G'S. to such a degree that the match lies on the other
side of the second point of intersection of the red and
green sensation curves, viz. at SSN. 23-6. (When the
factor of fatigue is much smaller, the fatigue colour has
to be closer to SSN. 236 before the match lies on the
other side.)

Treating all these fatigue colours and these matches
as we have done in this observation, it will be found
that when the white has been deducted from their com-
position, after the fatigue factor has been applied to the
former, the residue in each will be either R'S. + B'S. or
G'S. + B'S., and never R'S. + B'S. in one and G'S. + B'S.
in the other.

Fatigue Matches with smaller Brightness of Red,

A record of observations made with the fatiguing
red reduced to closely one-sixth of that used in the last
record. Table LXIX., is shown in the next table. The

378 RESEARCHES IN COLOUR VISION

factor of red fatigue is calculated in the same way as
before. Alongside is another record where the &tigue
was induced by a patch of ** red lithium" light reflected
from a card which had a luminosity for the D light of 2
candles at 1 ft. distant from the screen. This makes
the red light equal to closely one-fifth of a candle. The
factors derived from this record are also given.

Table LXX.

Record of Observation with the Eye

Record of Obaenration when the Eve

Fatigued by "

Red Lithium

was Fatigued with " Red Lithium Light."

Light," about one-sixth that given in

of a Luminosity of about one-fifth candle

Table LXIX.

at 1 ft. from Screen.

• C^

•s

^

50 .

^^

r&S

«5 .

Fatigued E
observed 88

•si

No

Factor of B
calculated

Remarks.

• ••

Fatigued E
observed S^

Unfatigued ]
Matched with

Factor of R
calculated

Remarks.

58-6

• > •

■ • ■

• • •

• • ■

1

chnnjife

50

54-4

•58

• ■ •

66

* • •

■ • ■

«••

53-3

52

-63

• • ■

53-3

52-8

•81

« • •

, 507

48 1

•54

• • •

507

50-2

-87

a ■ a

48-6

44 3

•61

48-6

47-4

-82

• > •

45 4

40

•59

■ ■ •

45-4

42-8

•82

■ • *

42 8

39 1

•61

Mean factor '61.

42-8

39-8

•8

• ••

40 1

35-9

* • •

Beyond 42*8 no

■ • «

• ••

« • •

• • •

37-5

35-9

» • •

factor could be
calculated , an pro-

37-5

36-4

75

Mean factor for
R'S. 81

34-9

3.13

• •

portion of G'S. to
K'S. was impos-

« • •

• • •

• • •

« V •

32-2

No

32-2

334

■ • ■

Each reading
seemed good

1

change

sible, being

And no

1

smaller than that

change

, 29-6

327

^ven in Table

• • •

■ ■ •

« ■ •

• •»

. 27

291

■ ■ ■

LXVIII.

27

28-5

■ • •

■ «•

1 21-7

27-2

...

■ • ■

217

28

• « ■

• • •

16-5

28

• • •

■ ■ ■

11-2

217

• ft *

The violet becomes
blue

In the record of matches given, the lowest naimber
recorded is SSN. 16* 5. Below this the matches could
not be made with the highest factor of fatigue ("34)
given, as the colour disappeared entirely from view to

ON COLOUR FATIGUE 379

the fatigued eye, though still perfectly visible to the eye
which was unfatigued. It was curious and interesting
to see the colour (which was always much bluer) gradu-
ally appearing.

(When the retina is fatigued by a colour, there is
always a certain amount of general insensitiveness in-
duced, and what Burch calls dazzle effect, as measure-
ments have shown, in several parts of. the spectrum. A
reference to pages 186-7 will give the probable reason
of this, for although there only an example of the effect
of the illumination of the retina by white light is shown,
yet the same applies when the retinal illumination is a
spectrum colour.)

At SSN. 10, at which point there are only red and
blue sensations present, the green being absent, the
violet became a dark blue, showing the partial oblitera-
tion of the R'S.

Matching the Spect7*uni Colours when Fatigued

by Green,

An example of fatigue with a green ^ ray will now
be recorded, the ray being that which has equal amounts
of R'S. and B'S. and the large excess being G'S. Such a
ray has been discussed on pp. 370-1.

The matches for this green fatigue were made by
another observer independently. A large number of
fatigue observations have been made by him, as also by
the writer. The factors derived from the measures are
given in the third column, and, as in Table LXX., the
measures of a diminished fatigue caused by looking at a
card at M illuminated by this same colour is given.

^ It may be said that to the aathor the constancy of the fatigue by the
red ray is more readily obtained than by the green.

380 RESEARCHES IN COLOUR VISION

Table LXXI.

Record of Observation with the Retina

R«cord of Obserratioo with tbe Retina

Fatigued with

a Green Ray at

Fatigaed with a Green Ray at

aSN. 36-6 if the Luminosity of the

88N.41-6.

D Light=one-fifth candle at 1 ft from

the Card Screen.

^

?>-.i

^

•

Is

5^^-

1

r

•1^

II

21

2^ ! Remarks.

21

OS

Remarks.

-5 S ^•^

ss

:>4

u

II

<

s

0S.

68-6

68*6

• « •

No measurable
change.

58'6

58-6

• « •

...

1

56

56

• •*

No match, fatigue
pink.

56

56

■ ■•

No measurable |
change.

53*3

54*6

•68

Fatigue still rather
pink.

53-3

54-5

'63

■ • ■

60-7

52-4

'63

Fatigue a little blue.

607

51-36

•87

• • •

46-6

60-9

'68

(Jood match.

48-6

4975

78

• • «

464

46-8

-60

• • •

45-4

47

•81

* • •

42-8

46-6

'46

• « •

42-8

5115

•80

■ • ■

401

447

•67

• • •

• ■ •

« • •

• • ■

• • •

87-6

41

•66

•Tlie amount of RU

37-5

38-6

•82

...

34 T*

36

is the smallest and

• • •

• ■ •

• • >

32*2

29*4

all of it is taken to

32-2

33-6

See Remarks.*

27t

26

form white. Only

27

23-1

Paler.

21 7t

22t

green and blue re-

21-7

217

No change.

16 •5t

16-6

main with the

white.
fUnfatigued colour

▼ery much paler.
^Practically no

11-2

11-2

No change, but
darker (the 6'S.
was nil, and
there was only

fatigue of the R'S.

change.

andB'S.).

By this method of matchiug we have established the
fact that we can find the factor of fatigue for the red
and green sensations. If the factor of fatigue is the
same as the factor of the sensations as described in pre-
ceding chapters, of the incompletely colour blind, then
the luminosity curves measured with a fatigued eye
ought to obey the same rule as those measured by the
colour blind.

ON COLOUR FATIGUE

381

Luminosity Curves {of Equal Areas) of the Colour Blind
pass through one point in the Normal Luminosity
Curve.

Dr. Watson has shown that theoretically the lumi-
nosity curves of all degrees of colour blindness, and with
approximately the same degree of "yellow spot" pig-
mentation when reduced to equal areas, ought to cut the
normal curve at a point where the ordinate is that of
SSN. 48 '6, the place in the spectrum where the R'S. and
G'S. of the "equal sensation area" curves cut.

Table LXXII. — Luminosity Curves all reduced to Equal Areas

{see page for Luminosities),

SSN.

■

Completely Green Blind.

Completely Bed Blind.

Im

62

o
S5

E.

2-8

F.

D.

G.

H.

K.

L.
...

X.

2

2-8

2-8

■ • •

• ■ ■

■ • •

■ • •

60

7

9-9

10-7

101

• ■ •

■ • •

a ■ •

...

■ • •

58

21

28-2

30-8

31-2

• • •

• • •

• • •

...

•7

56

60

64-8

67-8

64-8

11-3

10*3

9-4

10-1

7-4

54

80

105-7

1005

101-4

27-7

25-7

23-8

23-5

24-7

52

96

112-7

115-7

113-3

56-7

56-5

45-9

57-1

49-8

50

100

105-7

108-8

104-7

92-2

82-3

87-8

83-9

82-6

48

97

94-5

91-6

93

107-2

102-8

107-2

100-8

105-7

46

87

78-9

79

78

112

111-3

110-6

110-8

108

44

75

63-4

65

64

104

109-7

104-6

107-4

104

42

62-5

50

49-4

49-8

88-8

95-9

94-7

94

90-7

40

50

38-2

37-3

37-3

73-1

83-9

78

77-3

79-2

38

36

23-3

25-4

25-3

51-9

58-2

60-6

57-1

611

36

24

14-1

15-5

16-7

32-4

39-4

41-8

33-6

42-9

34

14-2

8-7

8-8

10-1

20-5

22-3

22-8

22-8

231

32

8-5

6-3

4-8

7

13

13-7

14-7

14-8

14-9

30

5-7

4-8

3-5

51

8-2

9-6

8-4

10-7

9-9

28

4

42

2-8

3-9

7-5

6-9

6-7

8-7

6-6

26

2-8

3-5

2-3

2-9

6-5

3-4

4

6-4

4

24

1-9

2-8

1-8

2-1

4-8

2-4

2-7

3-4

31

22

1-4

2-4

1-4

1-7

3-8

1-7

1-7

2-4

2-6

20

11

2-1

1-1

1-4

2-7

1

1

1

1-8

382 RESEARCHES IN COLOUR VISION

Table LXXII. shows the curves given in Chapter XIX.
of the observers who had complete red or green blind-
ness made of equal areas. Table LXXIII. shows the
curves of some of the incompletely colour blind persons
mentioned in Chapter XX. and the following chap-
ters, also made of equal areas, and also those of
completely colour blind ('5 R'S. and -6 G'S.) calculated
from Table XXXVIII. All these curves cut in the
place named above.

Table JjXXIU.^ Partially Colour Blind from Chapters XX,,

XXI.y ^c. reditced to Eqtud Areas.

•

'3'S

Incomplete B1indne».

88X.

1 |i

W.X116

N.xl-37.

N.W.x

112.

Z.Xl'32.

Si

•J

JG'8.

i R'S.
1-23

oG-a

OE-S.

62

2

• • •

• • •

• • •

• • •

26

2-8

60

7

2-9

9-8

33

4-6

8-7

4

10

... j

a **

68

21

9-1

28-6

11-2

15-8

26-4

13-4

29-3

•7

56

50

23

65-8

33-6

35-6

60-8

33-7

68

7-6

54

80

48-9

99

68-2

62

93-6

69-5

104

26-5

52

96

72-4

112-3

78-4

81-8

107-3

79

117-8

61-5

50

100

94-9

105-6

90-7

96-4

104-6

93-6

107-2

83-9

48

97

106-4

92-7

97-4

101-6

94-9

100

94

106-5

46

87

106-4

78-2

100-8

97-7

81-2

94-5

78

109-6

44

75

97-7

63-6

96-6

88-4

67-6

86-7

63-2

1031

42

62-5

84

60-6

80-6

75-2

54-9

73-6

49-6

93

40

50

71-3

39-2

70

62

42-4

61-7

37-2

80-7

38

36

54-6

26-6

57-5

46*7

29-6

46-4

25-4

61

36

24

36-8

16-8

25-7

34-8

19-4

30

15-8

42-7

34

14-2

23

10*5

14

20

111

18-7

8-9

21-7

32

8-5

13-8

5-4

11-2

10-6

6-9

11-4

5-4

16

30

1

5-7

9-2

4

5-6

6-7

4-35

7-6

4-4

10-2

Fig. 99 shows the luminosity curves of the normal
eye, completely green and red blind, and of persons
having only ^ R'S. and J G'S., and shows how com-
pletely accurate the intersections are.

If fatigue induces temporary colour blindness, then

ON COLOUR FATIGUE 383

the luminosity curves taken with the fiitigued eye should
also cut the normal curve at the ordinate of SSN. 48'6.

Method of obtaining a Luminosity Curve with a
Fatigued Eye,

To obtain the fatigue luminosity curves, a rather
different arrangement had to be made from that of

Fig. 97. Much of the description of that figure applies
to Fig. 100. Only one screen is required.

The spectrum through A remains unaltered, hut no
third spectrum is required. The white light is used as

384 RESEARCHES IN COLOUR VISION

in ordinary luminosity measures, falling on C is used as a
comparison light, a rod, R, being placed in the path of
the beam from A and the white light to form a patch of
colour and white side by side. In the white beam an
annulus is placed to reduce the luminosity of the white
(see p. 40). The fatigue is caused by the ray from the

r-^

MV.*:''

B r

l--^^'

Fig. 100.

second spectrum B, which has two mirrors M' and M
placed in its path to reflect it to M and thence to E'^.

//

Luminosity Curves obtained with a Fatigued

Retina.

The measures in Table LXXIV. are those of the
luminosity of a spectrum of which the source of light was
the crater of the positive pole of the electric light, the
positive pole being horizontal. (Table XL., p. 244,
should be consulted.) The fatigue was induced by
"red lithium" light.

S I-

ON COLOUR FATIGUE

385

\ <r

Tablb LXXIV. — Luminoeity of the Spectrum measured — (1) by an
Unfaiifjued Eye] (2) hy an Eye Fatigued with Red Light at the
place of the Red Litfiium.

o.^

o o

^

SSN.

1

•Bo

Luminosi
Fatigued

Unfatigued
Colour.

White.

Fatigue Colour.

Fatigue White.

59-8

10

8

Red

White

Red

Pale green

1 58-6

17-3

14

Red

1 1

II

57-3

32-2

25

Very red scarlet

1 1

Reddish

56

48-3

44

Red scarlet

II

II

54-6

66

58-5

Scarlet

1 1

Pale yellow

53-3

73

70

Orange

II

Yellow-orange

52

84-5

80

II

1 1

Pale yellow

507

92

90

II

II

Bright yellow

48-6

100

100

Greenish yellow

Pinkish

Green

46-7

98

100-5

Yellow-green

ii

• 1

1 45-4

92

96

Green

1

1 1

44-2

87

92

II

1

Pale green

42-8

78

81-5

Pure green

1

Very pale green

Very pale blue

41-4

67

73

II

1

II

1 1

401

58

60-5

Green

1

Paleslate colour

Green

88-8

48

52

Full green

II

II

II

37-6

38

43

Blue-green

1

Verjr pale

Very pale g^een

1

1 «

bluish green

1 361

28

83

Slight green-
blue

II

Dull grey

Very pale bluish
green

34-9

22

24-5

Blue-green

Yellowish

II

Pale bluish

33-6

14

16

Greenish blue

• 1

Grey

green
1 1

82-2

9-5

11

Blue

Dirty yellow

Blue

Very pale green

29-6

5

8-5

! 1.

Yellow

Strong blue

•I

27

4-2

5-5

• •

II

II

II

24-4

2-8

6-5

Reddish blue

Amber

Blue

Amber

21-7

2-1

4*8

I Nearly purple

1 1

1 1

Greenish yellow

16-5

1-84

27

Purple

• 1

1 •

Buff

The unfatigued colours and white, and also the
fatigued colours and white, are given. When these
numbers are graphically shown, the following are the
luminosities at whole numbers of the scale.

2b

386 RESEAECHES IN COLOUR VISION

1

Iablb T.XXV.

SSN.

Unfatigned.

Fatigued.

60

8*7

8

58

215

19

1 56

48-3

44

1 54

70

64

52

84-7

80

1 50

962

94

48

99-9

101

46

95

99

44

85-3

89

42

72

77

40

561

60

38

41

45

36

27-5

30-5

34

15-8

19

32

8-9

10-5

30

62

85

28

4*6

7-3

26

35

6-1

24

2-7

5-2

22

2*2

43

20

1-76

3-5

•

18

1-48

2-8

16 .

1-29

2-2

•

In Table LXXIV. the measurements of luminosity at
487 were the same for the unfatigued eye as that for
the fatigued eye on the annulus scale, and show 100.
When making the observations the luminosity measures
were made with the eyes unfatigued. The eye was
then fatigued by the red, and when the fatigue
and recovery were a balance the mirror M was
sharply turned against K. (Fig. 99) and the white
altered till the luminosities appeared equal. Three
or four measures were made and the mean taken as
correct.

We can now apply Method I., Chapter XX., for
obtaining the amount of fatigue of the red sensation.

ON COLOUR FATIGUE 387

Taking two numbers widely apart, viz. SSN.'s 56
and 36, we form the two necessary equations —

SSN. 56 is 48'3-44z=46-ly

SSN. 36 is 27-5-30-5z=12-7y

These equations make y='303, 2 = '776, ar=*697 as the
factor for R'S. {y being the deficit ; see p. 299).
Again, from SSN.'s 56 and 46 we get —

48-3-44 = 46-l
95-99 = 59
y=-326, 2 = 75, 0.-= -674
SSN.'s 54 and 44 give—

70-642=68-3y
85-3-89 = 497.y
y=-316, 2 = 782, a; = -684
Taking SSN.'s 50 and 42, we get from the equations —

y=-329, 2 = 766, ar = 67l
SSN.'s 58 and 38 give—

7/= -325, ;r=-675, 2 = 768
SSN.'s 50 and 40 give—

y=-313, a:=-687, z=783

The mean value of x is "68 and of z 77.

The factor of R'S. fatigue has been found by
exactly the same method as was done for the "colour,
blind factor." The luminosity curves of the colour
blind eye, and of this fatigued eye cut the normal at
SSN. 48-6.

Another example of the luminosity curve taken by
an eye rather strongly fatigued by the green at SSN.
37*5 is given.

388 RESEARCHES IN COLOUR VISION

Table ULXYl.^Faiigued with Green, Ray at SSX. 375.

LDminaritiefl.

Unfatigned.

. 8SN.

t

fatigued.

10
17-3
48*3
73

84*5

Patigiied.

09-8

66*6

66

63*8

607

26-3
64
92
103

48-6

100

100

46*4

92

84

42*8

78

63

401

58

46

37*5

38

28

34-9

22

16

32*2

9-5

8*5

29-6

57

61

Colour.

Red

Scarlet
Orange
Yellowish

orange
Greenish

Tellow
YeUow-green

Oreen

f»

White.

White

ft

Pinkish

white

Pinkish

tt

Pinkish
white
Blue-green Yellowish

»f

Greenish

blue
Blue

Dirty

yellow
Yellow

Fatigued.

Colour.

Red

Scariet

Reddish
yellow

Very pale
yellow

Yellowish
white

Dusty green-
ish white

Bluish green

Very pale

blue'green
Pale blue-

Pale blue

li
Blue

White

Pink

Wliite

tt

Very pale

maure pink
Strawberry ■
cream

t>

It

Pinkish

Pink

From this table the following curve was made

Table LXXVII.

SSN.

Unfatigued.

Fatigued.
11-5

60

87

58

21-5

34

66

48-3

64

54

70

86

52

84-7

100

60

96-2

103

48

99-9

97

46

95

86

44

85-3

76

42

72

60

40

66-1

45

38

41

30

36

27*5

20

34

15-8

11-5

32

8-9

6

30

6-2

6

ON COLOUR FATIGUE 389

Treating their luminosity in the same way as in the
last table, the following factors were found : —

From SSN.'s 56 and 36—

y = -87,x='13, z = 735
From SSN.'s 52 and 38—

y = -8S,x=17, z = -737
From SSN.'s 52 and 40—

2/= -86, «= -14, 5:= -734
From SSN.'s 52 and 34—

y=-85, a: = -15, 2; = -732

From SSN.'s 56 and 36—

y=-85, .r=-15, 2=735
From SSN.'s 54 and 44—

y=-85, a:=-15, 2=760

The mean is closely '15 for ^, the factor of fatigue,
and the mean value of z is 74.

Numerous measures of the luminosity curves of the
spectrum with fatigue by diflTerent colours and by
different intensities have been made, some of which
give more regular factors than others, but agreeing to
the first place of decimals. To get a steady fatigue and
to make a rapid observation of the luminosity are all
that is required to give equally good readings as can be
made with the unfatigued eye. This part of the fatigue
observations is now left, and we shall consider the results
of white and violet fatigue.

390 RESEARCHES IN COLOUR VISION

Colours of the Spectrum when Fatigued by White.

We have already stated that (Fig. 98) white is pro-
duced when all three sensations are equally stimulated.
The fatigue of the retina by white therefore ought to be
instructive. It should confirm the existence of white
in a large part of the spectrum as shown in Fig. 84. If
some colours of the spectrum are mixed with white, we
should expect that, if the white is partially destroyed,
the colours should become less yellow {i.e. bluer), and
this would be indicated by the match with an eye fatigued
with white.

Table I.X.1LYU1.^ Faiigued tcith }Vk{te.

Fatigued
Eye

8SN.

68-6

66

53-3

60-7
48-6
45-4
42-8
40-1
37-6
34-9
32 2
29-6
27

21-7

16-5

Match

by Un-

fatigued

Eye

SSN.

58-6

56

63*3

50-2
48-6
44-9
40-7
38-1
33-6
31-2
27-6
24-9
27

21-7

16-5

Remarks.

The match colour was slightly paler.

»>

>»

tt

No change which could be made ; match colour a little

pinker.
Match nearly perfect.

No change ; the match colour a shade paler.
The match colour was a little paler.

»»

»>

i»

ft

jj

The match was pale.

Good match.

Match colour paler.

The fatigue colour invisible at first ; match paler.

Fatigue colour invisible at first ; match a little paler

tbAn the fatigue colour.
No change in hue, but the fatigue colour was invisible

at first and the match colour pale blue.
No chance in hue, but the fatigue colour was a deep

purplisn blue ; the match colour was a pale purple.

Taking the red end of the spectrum from SSN/s 58*6 to
53*3, we find that the matching colour was slightly pinker

ON COLOUR FATIGUE 391

in hue, but no change in the spectrum colours could be
found. The tinge of pink shows that there is a certain
very small quantity of white in this part of the spec-
trum, due either to an immeasurably small quantity of
the green and blue sensations being present, or else to
the white due to the illumination of the prisms of the
apparatus. (A very minute quantity of white added to
the spectrum red makes it slightly pink. ) The pink is
not observed ordinarily, but only in contrast with the
fatigued colour. At SSN. 50*7 there is a definite shift of
the match colour towards SSN. 48*7, the point at which
there is no change in hue by the addition of white (see
Chapter XVII.). After SSN. 487 the " match " colours
always have lower scale numbers than the "fatigue"
colours. The matches are therefore all bluer than the
fatigue colours — that is, the absence of part of the white
makes the fatigue colour less yellow than it is with the
normal white present. At SSN. 27 the fatigue gives us
an impression as to what the colour of the blue sensa-
tion is when it is mixed with a comparatively small
quantity of white which the fatigue largely obliterated.
It seems to be a reddish blue, a colour which near the
place of the blue lithium line in the spectrum is, from
the sensation curves, a colour which represents the blue
sensation, but being mixed with about 80 per cent, of
white is not recognised as the colour of the sensation.

Spectrum Colours Fatigued by Violet.

Table LXXIX. shows the fatigue by violet at SSN.
8, in which only red sensations and blue sensations are
to be found. The fatigue of R'S. compared with that
of B'S. (see Fig. 98) is very small, K'S. 2*2 to B'S,
97*8, and the former can hardly be expected to show.

392 RESEARCHES IN COLOUR VISION

On the other hand, the fatigue by the B'S. will be much
larger.

Table IjXXIX. ^Fatigue with Violet at SSN, 8.

FaUgued Eye 8SN.

58-6

56

53*3

60-7

48-6

45-4

42-8

40-1

37-5

34-9

32-2

29-6

27
21-7

16-5

Match by Unfaiigued i
Eye S8N.

Reniarics.

68-6^

56

63-3

50-7

48-6J

47

45-4

44-2

42

43

40-7

42-8

36-9
21-7

5-1

The colours to the unfatigned
eye look yellower than to the
' fatigued eye, but no match can
be accurately made.

Match much paler.

it

It

Match still whiter than last.

if

little

seems nearly white.

pale yellowish white.

very white, with a

yellow.

Match nearly white.

With unfatigued eye paler than

with the fatigued eye.
Fatigue colour invisible at first,
but very pale, probably owing to
feeble luminosity.

From SSN/s 58*6 to 487 there is no apparent change ;
the fatigue colour looks a little yellower. This small
change in hue it was not possible to match. The G'S.
and R'S. mixed with the remnant of blue existing in this
region, made the fatigue colour yellower, the reason
being the same as that given in considering the " white "
fatigue. Taking the remarks as to 487, it says the
match of the unfatigued colour is paler than the fatigued
colour. This is to be expected, as only a part of the
B'S. has been left in the fatigued retina.

After leaving this SSN. and passing towards the blue
end of the spectrum, we have a different state of the
sensations. The blue sensation commences to increase
rapidly, and the effect of the fatigue is therefore more
marked. This the table shows. The fatigue of the B'S.

.=L=:i

ON COLOUK FATIGUE 393

diminishes the white present, and therefore the colours
are purer than seen in the spectrum. As the addition
of white makes a colour yellow, the abstraction of white,
as before, makes it bluer. This is the case till SSN.
21 '7, when there is no measurable change, but the
"unfatigued" colour is paler than the "fatigued," as
could be predicted from the diagram. At SSN. 16-7 the
match is found in the pure violet, though it is imperfect
and is paler than the fatigue colour, which may be due
to the luminosity falling nearly into that which deprives
it of any colour.

Many more places in the spectrum were taken as
fatiguing colours, and the results show, as stated, that
the fatigue factors by all can be classed as red or green
incomplete blindness, with the exception of SSN. 48*6.
This gives no factor for those colours where all the
sensations are found.

The Addition of the Fatiguing Colour to the Match
Colours of the Fatigued Colours.

It has already been stated that no "sympathetic
action" has been found between the fatigued and
unfatigued eyes. A direct proof that this is the case
is to be found in the results of experiments made with
the "matching" colours and unfatigued eyes. When
the factor of fatigue for red or green has been found,
the addition of the necessary amount of the fatiguing
colour to the matching colours will give the hue of
the "fatigue" colour before it has been observed with
the fatigued eye. Thus, if the fatigue factor of the
red is, say, '3, the addition of '3 of the red sensation
existing in the "fatigue" colour to the match colour
when the luminosities are the same, will give the hue

394 RESEARCHES IN COLOUR VISION

of the former colour before it was matched with
the fatigued eye. The same is the case when
the fatigue is caused by the green or violet. The
easiest plan of procedure is to add the necessary
amount of the fatiguing colour to the match to cause
its hue to be the same as that of the "unfatigued"
colour, and note the addition made. It will be
found that the amount added is that necessitated by
the factor of fatigue.

Colours of the Spectrum to the NovTiial Eye.

Below are tables of the colours at the different scale
numbers in the foregoing tables of this chapter as they
appear to the normal eye when not contrasted with
white or any other colour— (1st) when unfatigued ;
(2nd) when fatigued.

Table LXXX. — Unfatiffued.

Rather bluer green.

Bluish green.

Greenish blue (viridian).

Blue (pale).

Darker blue.

Reddish blue (not much

red).
Purple.

Violet purple.

Violet.

Violet.

SSN.

SSN.

61-2 .

. R«d.

401

68-6 .

. Red.

37-6

56

. Reddish scarlet.

34-9

64-4 .

. Scarlet.

32-2

53-3 .

. Red orange.

27

62-3 .

Orange.

21-6

51-2 .

. Yellow orange.

50-2 .

. Yellow.

19-2

491 .

. Greenish yellow.

16-6

481 .

. Yellowish green.

11-2

45-4 .

. Green.

8-2

42-8 .

. Slightly blue-green.

Table LXXXI. shows the colours when the eye
has been fatigued— (1) by red lithium red; and (2) by
green at SSN. 37 '5. The factors of fatigue for both
were closely '5.

ON COLOUR FATIGUE

395

Table LXXXI.

SSN.
69-8

Colours Fatigued with Red.
Ked much darker

Colours Fatigued with Green.

Bluer and darker

68-6

„ darker, but pale

Redder than unfatigued colour

56

Still pale and yellower

>» » ff i9

63-3

Pale yellow-green

, »» , » it }t

60-7

jf ff

Pinkish orange
Pale blue-pink

48-6

Bright yellowish green

45-4

" SoHd " green

„ bluer green

Dirty white

42-8

»> »j

401

ff }i )f

White slightly blue

37-6

Blue-green rather paler

» w »>

34-9

„ „ pale

„ rather bluer

32-2

Greenish blue same colour as
before fatigue, but paler

Pale blue

27

Saturated light blue

„ violet

21-7

„ blue

Dark purple (when the colour
appears)

16-6

,j darker blue

Very dark purple (when the
colour appears)

11-3

V » w

Very dark purple (when the
colour appears)

61

Deep blue

White

A slightly pale blue-green

A rather pale purple

These tables are given, as they will in some measure
aid the reader to understand to some extent why-
names are given to the colours of the spectrum by
colour blind persons with varying degrees of colour
blindness, which differ from those given by normal
vision. In the next chapter it will be seen how great are
the differences. By translating the colours of the above
table into the language of the colour blind (the founda-
tion of which is the colour they recognise as white),
a better idea will be gained of the reason for the
difference in nomenclature.

Rapid Fatigue.

Bidwell found that not only did fatigue take place
after a prolonged gaze at a colour, but that under certain

396 RESEARCHES IN COLOUR VISION

cooditloDs it took place after a momentary glance, and
was sufficient to produce a fugitive after image of the
complementary colour. The condition which governs
this is, that the retina shall be darkened imjnediately
before the bright object is viewed. He says: "The
retinal nerves, when in darkness, rapidly acquire a state
of sensitiveness far exceeding the normal average in the
light, but quickly diminishing again under the influence
of illumination. The peculiar sensitiveness may indeed
be both gained and lost in a small portion of a second,
and is therefore very fiivourable for the rapid generation
of negative after images."

Bidwell described methods by which this pheno-
menon can be shown in a very simple way. By using
a balanced disc rotating on a centre, and covering one-
half with black velvet and the other with white paper,
and at one junction cutting out a sector some 30® in
aperture, he was able to show it with great certainty.
A red wafer placed on a piece of grey paper is illu-
minated, and the disc is so placed that the black covers
it from view ; when the aperture rapidly passes in front
of the eye, so that the wafer is seen for an instant
through the opening, and the white of the disc shuts
it off, immediately there will be perceived a bright but
quickly-vanishing greenish-blue image. Bidwell states,
and the writer has confirmed it, that if the illumination
be strong and the disc rotated with proper speed, no
trace of red will be seen at all, and that the only
colour seen will be the blue-green after image. A
glance at Bid well's diagram, Fig. 8, p. 25, may throw
light on the subject of this rapid succession of nega-
tive after images, remembering that instead of white
we are using coloured light, and that the short dark
interval will now only be dark in respect to the

ON COLOUE FATIGUE 397

colour used, and not to the other colours. At a soiree
of the Royal Society, Bidwell showed a striking
example of this method of viewing after images. He
had a picture of a lady in which the hair was indigo
blue, an emerald green face, and a scarlet gown, who
was represented admiring a sunflower with purple
leaves. Seen through the disc revolving some six to
eight times a second, the hair appeared flaxen, the
face a delicate pink, and the dress peacock blue; the
sunflower became yellow, and its foliage green.

In Chapter VIII. is described an apparatus for
measuring " flicker " luminosity. By a slight alteration
of the apparatus a colour can be thrown on the small
square and then be followed by an interval of white light,
and this by an interval of darkness, and then an interval
of the colour again, and so on. This gives striking
after images of the colours, when the revolutions of the
flicker wheel is from 60 to 120 revolutions a minute.

CHAPTER XXVI

TESTING FOR COLOUR BLINDNESS

The preceding chapters will have given an idea of the
mode which has to be adopted in the laboratory for the
testing a person as to his colour vision when the eye
is practically dark adapted, and a further reference will
be made to certain of the tests therein indicated. The
question that now requires consideration is as to tests
which are available for the same purpose in daylight
where no laboratory is available. The acuteness of sensa-
tion for colour is not the same when the eye is saturated
with external white light as it is when it has been in dark-
ness for some time. The reason of this is perfectly easy
to understand. The three sensations are being per-
petually equally fatigued with white light, and they
consequently become less sensitive to any colour which
falls upon the retina. In making tests we have to
avoid being led astray by a person's ignorance of colour
nomenclature, though the writer has never met with
a case in which colour ignorance was at all pronounced.

The Wool Test

Holmgren, the Swedish scientist who has given in
the past much attention to the subject, recommended
that coloured skeins of wool should be employed to
diagnose colour blindness. This is based on the ground
that a variety of hues and shades of colours in that
form could be obtained in commerce. The wools never

898

TESTING FOE COLOUR BLINDNESS 399

appear **shiney," but reflect the light fairly equally in
every direction, and they are easily handled. He also
made it a sine qud non that the examinee should not be
called upon to name colours, but should be merely tested
by his picking out those skeins which fairly matched,
in colour only, certain standard colours that he selected
as efficient. The three colours which he selected were —

A pale green, i.e. a green much diluted

with white,
A full pink, and
A bright red.

The " confusion " skeins he selected contained different
shades of pinks, mauves, violets, neutral tints, browns,
dark greens, greens, buffs, yellows, green-yellows, blues,
green-blues, and others.

The standard colours selected are most suited for the
detection of complete or nearly complete colour blind-
ness rather than for colour blindness which is incom-
plete and is small.

Take the case of a completely red blind person, when
he is told to match the pale green scale, he would (in
the language of normal vision) see no red in the white
with which the green was mixed, and would select,
besides the correct matches^ a pale yellow or a neutral
tint or a buff as a match in colour, for he would dis-
tinguish none of the red in the colours which is recog-
nised by the normal eye in them.

A person completely green blind would see no green
in the green skein, and all he would see would be his
own form of white, a purple. He also would select pale
yellows or buffs or neutral tints, and, in fact, would
make a match with almost any pale coloured skein,
amongst others those which contain a suspicion of red

400 RESEARCHES IN COLOUR VISION

to the normal eye, the reason being that his white is, in
the language of normal vision, a purple, and most of the
pale colours would appear pale purple to him, and there-
fore he would take them as matches to the green skein.

With an incomplete blindness to green, but which
was of a pronounced character, the selection would not
be so varied. We have seen at p. 265 that a cer-
tain amount of colour may be added to white without
being perceived. As the green in the pale green scale is
slight, the colour may to such an eye disappear in his
white and not be diluted. His white would be a very pale
purple, and only those coloured skeins which looked of
the same whiteness (to him) would be selected.

Taking up the skeins examined by the complete red
or green blind and comparing them, it might be very
difficult, from the selections made, to determine whether
the person was red or green blind.

The use of the next test skein, the pink, would at
once determine into which category to place him.

To the completely red blind the red in what is
pink to the normal eye is non-existent. He sees the
pink as a pale blue, with perhaps a little green in
it, which comes from the white, and he will match
the skein with any colours which contain blue, whether
they contain red or not, thus he will pick out the
various shades of pink as matches; they, of course,
are as blue to him as the test skein, and he will
pick out the mauves and violets, as also the blue
skeins.

The completely green blind will see the pink nearly
as the normal eye sees it, and will pick out the pink
skeins, but will not think of matching the blue or
violet or mauve skeins with it, but he will select
pale bluish greens, as he sees no green, but only the

TESTING FOE COLOUR BLINDNESS 401

pink in them. He may also pick out a white or
neutral coloured skein, together with dark brown
and dark greens, for the absence of green sensation
may make them the colour (though darker than
it) of the pink skein. The third skein is a dark
scarlet, and the completely red blind would not see
the red in the scarlet, but only the dark green.
This enables him to match the dark green, and might
also place a light green with the test skein. On the
other hand, the completely green blind (or nearly
completely) would fail to recognise the green in the
scarlet, and would select all the scarlet and red skeins,
also the brown skeins.

This last test is the weakest of the three test skeins,
and would very often fail to detect colour blindness
if used alone. If, instead of the dark scarlet a dark
brown skein be substituted, much greater facility is
given to the detection of even fairly slight colour blind-
ness. The green component then is stronger, and the
red blind will match not only dark green but also light
green with it. The green blind would be inclined to
pick out the browns and the reds.

Except for the fairly pronounced examples of in-
complete colour blindness, it is not uncommon for the
incomplete colour blind to pass these three tests with
but slight errors. In order to detect these cases more
surely, the addition of two other test colours besides
the brown suffices, and though they may only approxi-
mately tell the amount of colour blindness that exists,
yet they will indicate that there is colour blindness,
and also its nature. The first of these extra tests
is a bluish purple. In this colour there is but a
small quantity of red, so that the incompletely red

blind would fail to notice it, and would make matches

2c

402 RESEARCHES IN COLOUR VISION

of pure blue with it, and probably add blues with
a small quantity of green in them. Such matches
would indicate a deficiency in the red sensation.
The incompletely green blind would match the proper
skeins with it, and very likely make no mistakes.

The next test skein should be a slightly pale
yellow. A red blind, besides the correct matches, will
select skeins in which the green-yellow preponderates,
and this may be taken as a sure sign that there is a
red deficiency in his sensations. The green blind will,
besides the proper matches, pick out yellows which
are decidedly on the red side of the test skein hue.
Such matches show that there is a green sensation
deficiency in the eye. If the test is properly applied,
the chances that any notable deficiency in any of the
sensations will escape detection are very small. After
an examination has been made, if the examinee is
asked to name some of the confusion colours, the
giving of a wrong name to any of them will confirm
what has probably been found out by the matches.

Test by Water-colour Washes.

Another daylight test which the writer has carried
out in the absence of the wools is by means of his water-
colour paint-box. This is simply a test for those who
are not colour ignorant. Before us is a test that
was made abroad of the eyes of a certain foreigner,
who had no idea that he was or could be colour blind.
A piece of drawing-paper, a tumbler of water, the
paint-box, and daylight was all that was required.

Pale stripes of rose madder of neutral tint, of
viridian (bluish green), aureolin, permanent violet,
with vermilion and emerald green, were brushed on

TESTING FOR COLOUR BLINDNESS 403

to the drawing-paper, and, when dry, one by one the
stripes were submitted and the hue of the wash asked
for. The rose madder was named pink, the neutral
tint pinkish, the permanent violet correctly, viridian
was grey, and so was the emerald green, the yellow
aureolin was reddish. These answers were quite
enough to convict the examinee of being partially
green blind.

Subsequently the same colours were submitted to
another person, whose eyes were supposed to be normal.

The faint vermilion was called green, the rose
madder was '' bluish," the neutral tint was whitish,
the viridian was bluish, aureolin yellow was green,
and the permanent violet was blue. The examinee
was evidently partially red blind. By mastering the
principles which underlie the trichromatic theory, it
is easy to make tests by coloured materials other than
the wools.

In Plate III. the water-colour washes were as
follow : —

No.

1. Neutral tint.

2. Neutral tint and a little viridian.

3. „ „ „ „ vermiHon.

4. „ „ „ „ cobalt blue.
6. Pale neutral tint and a little viridian.

6. f, ,f ,y „ „ rose madder.

7. ,, y, „ ff „ rose madder and cobalt blue.

8. Pale madder yellow.

9. Pale cobalt blue.

10. Mixture of cobalt blue and viridian.

11. „ „ aureolin and cyanine blue.

12. „ „ vermilion and blue.

13. Neutral tint.

14. Aureolin.

15. Bose madder.

These, of course, can be supplemented ad lib, by washes such as palo
vermilion and Hooker's green, &c

404 RESEAECHES IN COLOUR VISION

Test hy Colour Discs.

A test which can be applied qualitatively as well
as quantitatively is that of the rotating colour discs
of red and green, with black and white sectors behind
the smaller pair (see p. 337). The examinee may
make a match in daylight looking through a chromate
cell, containing chromate of potash in solution. The
angle of the red or green is altered until the two
give a yellow which matches in hue the outside disc.
The angles of the sectors are noted.

When a colour blind is called upon to make a
match, a very small angle should be shown of one of
the discs. This will probably be demanded to be in-
creased, the inner disc being said to be either too red or
too green. When a match is made, the angles of the
discs should be noted and a rough estimate can be made
by a comparison of the normal equation with that of
the examinee.^ If the red sector is the greater, the
latter will be incompletely red blind ; and if the green
sector is the greater (compared with the normal), there
is incomplete green blindness. A small motor, of course,
is most useful for rotating the discs, but there are
several mechanical whirling apparatus which can be
employed.

Laboratory Tests.

In the laboratory with the colour patch apparatus
a very fair idea of the amount of deficiaicy in the red
and green sensations is given by noting the names given
to the colours at various parts of the spectrum. In
Chapter XVIII. specimens of the colours named by
completely and nearly completely red and green blind
have been illustrated, as also the alteration in nonien-

^ This is more fully described in Chapter XXIII.

PLATE /r.

Xoniiifl.

■75 KS.

•5 c;s

B J

T-H' r

1 us.

1 — ^ — t

•25 RS

T T

JUL

2 RS

^

25 GS.

Xonuiil

Sjx'ctrum O^h^urs as named by persons with different decrees

of Colour Blindness.

TESTING FOR COLOUR BLINDNESS 405

clature of the colours under diflfering conditions in the
case of a person who was very nearly completely but
not quite red blind. In Plate IV. we have illustrations
of the spectrum colours named by eight different persons,
with varying degrees of red or green blindness, when
the colours were shown in patches. One slit in the
spectrum was employed, and was placed at the extreme
limit of the red end of the spectrum (to the normal eye).
The examinee was asked to name the colour, and in the
cases of red blindness he usually saw nothing at the
extreme red but a faint colourless or green light, due
to the small quantity of white illuminating the prism.
The slit was gradually moved along the spectrum.^ The
examinee was required to stop the movement when he
saw any change in the red, and was asked to name it.
The slit was again moved in the spectrum till another
change in colour was seen, and this scale number was
noted. In this way the whole of the spectrum was
submitted for colour naming.

When the extreme violet was reached, the colours
were submitted with the motion of the slit reversed.
(Of course, as in the two previous plates, no attempt
has been made to indicate the luminosity, but only
the colour.) There are very marked differences in the
colours named by eyes possessing various degrees of
colour blindness.^ The amount of green or red blind-
ness was determined by the methods given in Chapter
XX. Again, if a patch of colour be shown alongside a
patch of white, those who are completely blind or
nearly completely blind to the sensation will match
them together when the colour is a green. The com-

* In the case of the red blind it was noted when the patch appeared red.

■ In one case ('2 RS.) above the patch of yellow are two red marks.
These marks indicate that on two separate occasions red was named at
this position instead of yellow.

406 RESEARCHES IN COLOUR VISION

pletely red blind will with the arc light make a match
about SSN/s 34 to 35, and the completely green blind
about SSN/s 36 to 38; the exact position will vary
a little, according to the amount of pigment present in
the macula lutea. This position is often called the
neutral point in the spectrum, and will be found when
examining completely colour blind and nearly completely
colour blind. In the cases of green blind who have a
small factor for green sensation, it may happen that
a large band of spectrum may be matched with the
white sometimes extending from the red to the blue
(see Plate II.). This is easily accounted for, owing to
a small amount of colour in a large amount of white
escaping detection (see Chapter XIX.). The case cited
on p. 300 is an example of this kind.

Dot Test.

Another test (called the dot test) is to throw the
different colours on a small white disc about ^ in. dia-
meter, mounted on black velvet, and make the examinee
stand at from 12 ft. to 20 ft. away, and name the colours
shown. Chapter XII. shows that by diminishing the
angular measurement of a patch of colour, it becomes
colourless.' As one of the colour sensations to the colour
blind is less than to normal vision, it follows that the
small patch may fail to show a colour to the colour blind
when it is quite visible to the normal eye. If the
examinee have a red sensation deficiency, the colours
in the red end of the spectrum may not be visible at
all, or else be nearly colourless. When the factor of

* It is shown that the intensity of a patch of light to first lose its colour
may be increased tenfold when the size of the patch is reduced to j^ in.
diameter.

PLATE V,

40

Dots of
pure colour

Pure colours
mixed with
white

Colour
contrasted
with white

Patches of
pure colour

Normal

Spectrum Colours as named by a person who possessed

•35 of Red Sensation.

TESTING FOR COLOUR BLINDNESS 407

the sensation is large, only the dark reds will fail to be
seen, though they are quite apparent to the normal vision.

Plate V. shows the striking differences in colour
nomenclature given by the same person when the
colour patches are shown under differing conditions,
such as those above mentioned.

Band No. 5 is the spectrum to the normal eye.

The remaining bands of colours apply to a colour
blind whose factor of red sensation was '35.

Band No. 4 were the colours named when patches
of colour only were thrown on the screen, and is com-
parable with Plate IV.

Band No. 3 shows the colours named when con-
trasted with a white patch. ^

No. 2 band shows the spectrum colours as named
by the colour blind person when mixed with a small
quantity of white light.

No. 1 band shows the names he gives to the
spectrum colours when reduced to dots of ^ in. diameter,
and observed 16 feet away from the screen.

Plate V. may be compared with Plate II. as
examples of the non-recognition of colour by the colour
blind, when the spectrum colours are reduced on the
screen to very small areas. It will be noticed in both
that the names of the colours are very often quite different
to those which are given when the patches are larger.
The variation is very well shown in Plate V. Here we
have colour blindness; and by keeping the intensity
constant and diminishing the diameter of the patch, the
colour is extinguished sooner than it is to normal vision.
The amount of reduction necessary to extinguish depends
on the factor of the red or green sensation,

* When a nonnal eye is fatigued with red so that the factor of red fatigue
was about the same as the above factor, the white band shown as such in
Nos. 2, 3, and 4 appeared green, and the white of the reflected white light
was also green.

408 RESEARCHES IN COLOUR VISION

Lantern Test.

The examination of seamen for colour blindness has
been largely carried out for the past twenty years by
the writer, and the dot test has formed a valuable aid
in ascertaining whether the colour vision was suflGiciently
good to enable the ordinary red and green lights of
ships at sea to be recognised at a distance. Recently
a Committee representing the physical and physio-
logical sciences and nautical experts have recommended
that the same kind of test should be carried out, but
with an oil lantern in a darkened room, in which the
eyes of the examinees become " dark adapted."

With this lantern ships' lights as seen at a distance
of 500 and 2000 yards can be imitated by illuminating
two circidar apertures of the necessary diameter with
an ordinary oil light, such as is used on board ship, and
testing the examinee with different variations of red,
green, and white* coming through the two apertures.
The glasses which give the coloured lights are the same
as those in actual use on board ship. The three lights
are made of about equal luminosity to the normal eye.
It has been found that with factors of red and green
blindness as high as '6 to '7 there are mistakes made
in naming the colours shown both in the spectrum dots
and in the lantern, the most common being in the
mistaking of white for red, and of green for white,
and of the white for green. In cases where the degree
of colour blindness is smaller, the red and the green in
the lantern are mistaken for one another. It must be
recollected that the glasses transmit impure colour, the
green light being largely contaminated with white,
and the red glass allows yellow rays to pass. This test

' Naked oil light.

TESTING FOR COLOUR BLINDNESS 409

is less severe, perhaps, than the test with the dots of
spectrum colours in one respect; but it has the great
advantage of being a practical test, and easily under-
stood by the ordinary person who has no views on
theories of colour vision.

The official report ^ which the above Committee issued
shows that tests of ordinary ships' lanterns were carried
out at Shoeburyness (and elsewhere), the lamps being
observed by the colour blind (who had been previously
tested in the laboratory) at different distances up to
3000 yards, and gave the same results as with the
lantern in the laboratory as to the factors of colour
blindness which could just not distinguish the colour
of the lights.

Optical Imperfection of the Eye causes inability

to recognise Coloured Light.

The writer, a few years ago, when considering other
causes than deficient colour sensation which might prevent
the recognition of colour, came to the conclusion that the
optical condition of the eye might be of such a nature
that small discs of coloured light might be taken as
colom^less or not seen at all. To confirm or disprove his
diagnosis, he made his eyes myopic, &c., and observed
ships' lights from the seacoast, and also known stars, and
found that with about half normal vision ships' lights at
2 miles were sometimes invisible or colourless, and that
only stars above the 4th or 5th magnitude could make
an impression on the retina. The reasons for such failure
which he gave to the above Committee were on the
following lines.

1 Report to the Departmental Committee on Sight Tests (Board of Trade),
presented to both Houses of Parliament; published by H.M. Stationery
Office, 1912.

♦ :/ liyJiKAHrHt.< IX COLOUR VISIOX

A |X/'r-t '/f ii^'Lt sacL as that coming from a
ril^^^f.t ^t;ar fona% an uosl&: oq the retii^ which is
a di^iC c^ a certaiii size, depending oo the diameter of
^^ih letiM ti tr^ eye ar>i its focal length- A lig^t at a
certain dUtance, though not a point, will fcMm a disc
image of the same diameter, and beyond that distance
t^ie diameter of trie disc will not vary, though it will
increaise within tliat limit. The coloored light will,
whilift diminishing in angular dimenjnons as it recedes
Yiey</nd this limit, still show <m the retina practically the
same size of disc, and the smaller light will be less
'^ i\H%ii^. " on tliat disc. (Even for normal eyes there is
a distance where the eye with normal vision would fail
to see the colour ; see Chapter XIL)

If the vision be less than normal, the disc formed
on the retina by a point of light will be larger than
with the normal eye, the diameter depending on the
amount of defect in the form vision. Thus the coloured
light is spread over a larger area than is the case with
normal form vision, and, consequently, say a green light,
which will appear to the latter as a disc of green of
decided colour, may be seen by the former as a
whitish patch, or, in the case of a red, may not be seen
at all.

In the Committee's report the result of the prac-
tical testing of eyes having less than normal form
vision for the recognition of ships' lights is given.
Th<5 distance at which there is a failure to recognise
c<jIour is alx)ut 2000 to 3000 yards when the vision is
half normal.

The laboratory lantern tests give the same results.

[It ifi, perhaps, to be regretted that in the practical
teHts carried out during the two fortnights (in the late
atitiunn and Bpring), no great variations in the clearness

I

TESTING FOR COLOUR BLINDNESS 411

of the atmosphere were met with, as the conditions
were all in favour of the colour blind and defective
form vision examinees.]

Test by Simultaneous Contrast.

Another instructive test is founded on the answers
given as to the colours found by the simultaneous contrast
between white and the different colours of the spectrum.
The contrasts which the colour blind see appear often
to be extraordinary, but when considered in the light of
the three-sensation theory they lose what may be called
their extravagance. The smaller the factor of the defi-
cient sensation, the more divergent from the normal do
the contrast colours of the white become. It is no
uncommon answer, for instance, when to the normal
eye a colour is green and the white is a salmon colour
that both stripes should be called green. The mistakes
made very readily indicate the nature and extent of the
colour blindness that is being examined. This, of course,
is only a qualitative test. As an example of the mis-
takes that may be made, the following contrasts were
observed by an eye which had only '5 of the normal
green sensation : —

Colour shown.

Normal Contrast

Colour to

Contrast of White

of White.
Green-grey

Colour Blind.
Green

to Colour Blind.
Red

Dark red

Brighter red

Green-grey

Red

Pale green

Orange

Blue

Green

Green

Greenish vellow

Umber colour •

Pale green

Blue

Pale green

La vend ar

Green

Red

Pure green

Pink

Blue-green

Red

Bluish green
Greenisn blue

Salmon

Dark green

Pale red

Yellow

Blue

Green

412 RESEARCHES IN COLOUR VISION

Quantitative Tests.

In the laboratory, of course, the quantitative tests
which have been indicated in Chapters XX. to XXIII.
can be employed. When testing by the difference in
luminosities of the normal and colour blind persons,
the flicker method of getting the luminosity is the
easiest plan to adopt, and gives correct results if both
normal and colour blind make observations by it. The
method of getting flicker luminosity has been described
in Chapter VIII.

If a diagram be made like that given in Fig. 99,
p. 383, showing three different degrees of red and green
colour blindness, say '7, '3 and 0, factors of red sensation,
calculation is very much shortened.

Making the luminosity of the normal and colour
blind the same at SSN. 487, the measures of the
luminosity of the colour blind can be applied to the
different scale numbers of the spectrum, and the ordinates
show between which of two curves the examinee's obser-
vation lies. A close approximation to the factor of the
sensation which is deficient can at once be made. When
the luminosities of several scale numbers are taken, they
should all give the same factor.

PAPERS BY THE AUTHOR REFERRED TO

IN THIS WORK

(1) "The Production of Monochromatic Light." PhiL Mag., Aug. 1886.

(2) "Colour Photometry." Phil. Tram. Roy. Soc., 1886.

(3) " Colour Photometry," Part II. PhiL Trans. Boy. Soc., 1888.

(4) "Colour Photometry," Part III. Phil. Tram. Roy. Soc., 1892.

(6) " The Colour Sensations in Terms of Luminosity." PhU. Tram. Roy.
Soc., 1899.

(6) " Modified Apparatus for the Measurement of Colour and its Applica-

tion to the Determination of Colour Sensations." Phil. Tram. Roy,
Soc., 1900.

(7) " The Sensitiveness of the Retina to Light and Colour." Phil. Tram.

Roy. Soc., 1897.

(8) " Transmission of Sunlight through the Earth's Atmosphere." Phil.

Tram. Roy. Soe., 1887.

(9) " Transmission of Sunlight through the Earth's Atmosphere," Part II.

Phil. Tram. Roy. Soc., 1892.

(10) *'The Measurement of the Luminosity and Intensity of Light re-

flected from Coloured Surfaces." PhU. Mag., 1889.

(11) "Measurement of Colour produced by Contrast." Proc. Roy, Soc.,

1894, vol. Ivi.

(12) " On the Limit of Visibility of the Different Rays of the Spectrum."

Proc. Roy. Soc., 1891, vol. xlix.

(13) "The Numerical Registration of Colour." Proc. Roy. Soc., 1891,

vol. xlix.

(14) "The Estimation of the Luminosity of Coloured Surfaces used for

Colour Discs." Proc. Roy, Soc., 1900, vol. Ixvii.
(16) " On the Colours of Sky Light, Sunlight, Cloud Light, and Candle
Light." Proc, Roy, Soc, 1893, vol. liv.

(16) "A Case of Monochromatic Vision," Proc, Roy. Soc,, 1900, vol. Ixvi.

(17) "On the Examination for Colour of Cases of Tobacco Scotoma and of

Abnormal Blindness." Proc. Roy. Soc., 1891, vol. xlix.

(18) "On Photographing Sources of Light with Monochromatic Rays."

Proc. Roy. Soc,, 1896, vol. Ix.

(19) "Effect of the Spectrum on Haloid Salts of Silver." Proc, Roy. Soc.,

1890, vol. xlvii.

(20) "Intensity of Radiation through Turbid Media." Proc R(yy, Soc,, 1886,

No. 244.

418

414 RESEARCHES IN COLOUR VISION

(21

(22

(23
(24
(25
(26
(27
(28
(20

''Colour Blindness and the Trichromatic Theory of Colour Vision."

Proe. Roy, Soe., A., vol. Ixxxiii., 1910.
"Colour Blindness and the Trichromatic Theory of Colour Vision,

Part II., Incomplete Bed or Green Blinchiess." Proc, Roy. Sac.,

A., vol. Izxxiv., 1910.
"Colour Blindness and the Trichromatic Theory of Colour Vision/'

Part III. Proc, Roy. Soc.f A., vol. Ixxxvii., 1911.
"On the Extinction of Colour by Reduction of Luminosity." Proc.

Roy. Soc., A., vol. Ixxxiii , 1910.
" On the Change of Hue of Spectrum Colours by Dilution with White

Light.'' Proc, Roy. Sbc., A., vol. Ixxxiii., 1909.
"Colour Blindness and the Trichromatic Theory of Colour Vision/'

Part IV. Proe. Roy. Soc., A., vol. IxxxviL, 1912.
** Extinction of Light by an Illuminated Retina." Proc. Roy. Soc.^ A.,

vol Ixxxvii., 1912.
" Trichromatic Theory : Measurement of Retinal Fatigue." Proc. Roy.

Soc., A., vol. Ixxxviii., 1912.
"Variation in Gradation of a developed Photographic Image when

impressed by Monochromatic Light of different Wave-lengths."

Proc, Roy, Soc., voL Ixviii., 1901, p. 300.

INDEX

Absorption and obstruction factors, 5
Absorption by transparent media and

by pigments, 76
Amyl - acetate lamp compared with

candle light, 190
Angalar dimensions of colour patch,

effect of, on coloar extinction, 165
Annulas, 70
Apparatus used in the extinction of

light on illuminated retina, 184
Arc lamps, 66, 66, 67
Arc spectrum (with horizontal positive

pole) sensation curves, 244
Ai^and burner, simultaneous contrast

by, 122

B. O.'s luminosity, 356

Benham's top, 28

Bidwell on rapid fatigue, 396

Bidwell's researches, 22-32

Blind spot, 16

Borders, coloured, to black lines, 26

Box nsed to measure the extinction of

colour, 148
Box used to measure the extinction of

light, 158
Burch on fatigue, 363

Ghabfentibb's law, 23

Chromate of potassium, intensity of
light transmitted by, 79

Collimator, 34

Colour, addition of, to white, 266

Colour blindness, dangers of, 271

Colour blindness, extent of, in popula-
tion, 269

Colour constants, 4

Colour fields, 190 et nq.

Colour patch apparatus, 33, 38, 44

Colours of patches of colours to the
normal eye, 394

Colours of spectrum of low intensity,
147

Colours, pure, 1

Colours, pure, by diffraction grating, 3

Colours, pure, by prisms, 2

Committee on sight tests, 408

Committee on sight tests, recommenda-
tion of, 409, 410

Comparison of flicker and shadow

luminosities, 110
Complementary and contrast colours,

112
Complementary colours, 113

Dalton's colour blindness, 267

Dangers of colour blindness, 271

Dark intervals between colours destroys
simultaneous contrast, 125

Dazzle colours, 220

Detection of amount of colour blind-
ness by colour equations, first method,
309-13 ; second method, 315

Determination of angles for colour
discs viewed in "chromate** light,
337

Diaphragms used in the extinction of
light, 160

Discs, colour, 130-43

Discs, colour, measurement of colour
blindness by, 329

Discs, colour, used for test of colour
blindness, 404

Discs, interlacing of, 133

Dominant wave-lengths, 129

Dot test, 406

Double reflection of a ray of light,
161

Blectromotor for blending colour in

colour discs, 132
Blectromotor for discs, 132
Emerald green, colour sensations in,

251
Bmerald green, reflection of spectrum

by, 79
Bquation, colour, solution of, second

method, 315
Bquations, colour, for spectrum colours,

238-40
Bvolution of colour sense, 18
Bvolution of the eye, 17
Bxtinction box, latest, 167
Extinction dependent on the least dia-
meter of aperture, 175
Extinction of colour and light, 144
Extinction of colour (diagrams), 151,

153

415

416 RESEARCHES IN COLOUR VISION

Sztinction of light by completely red

and green blind eye, 289-91
Extinction of light (curves for spectrum

colours), 164
Extinction of light of spectrum colours,

156
Extinction of light received exccntri-

cally on retina, 178
Extinction of light received on whole

retina, 164
Extinction of light received on yellow

spot, 164
Eye, diagrammatic, 9
Eye, sensitiveness of, when dark

adapted, 157
Eye, structure of, 8

Factors of fatigue, 372, 387, 389
Fatigue by excessively bright colours,

363
Fatigue by green, 380
Fatigue by red, 372-8
Fatigue by white, 367, 390
Fatigue colours of single patches, 395
Fatigue, examples of, 370
Fatigue of the retina, 360
Fatigue, qualitative observations of,

366
Fatigue, rapid, 396
Feeble spectrum, luminosity of, 96
Field, dependence of, on size of spot,

208
Fields, colour, apparatus for testing,

192
Fields, colour, change of intensity of

light and extent of field, 203
Fields, colour, measured by W.B., 199
Fields, diminution of, with reduction

in intensity, 204
Fields of impure or mixed colours, 200
Fields, similarity for different colours,

195
Fine particles, nature of, in the

atmosphere, 62
Finnigan and Moore's Benham top

experiments, 31, 32
Fixed points in the spectrum, 233
Flicker apparatus, 107
Flicker colours by fatigue, 397
Flicker luminosity, 106

Gas light, simultaneous contrast by,

123
Ghosts, 22
Gliisses used in the red and violet of

spectrum in the extinction of light,

163
Green sensation curve of a red blind,

282
Grey produced by red, green, and blue

discs, 136

Helmholtz's sensation curves, 214
Helmholtz's trichromatic theory, 213
Heredity and colour blindness, 269
Hue, change of, by addition of white,
255

Illuminated retina, examples of light

extinction by, 186
Illumioated retina, extinction of the
light in spectrum colours by an, 183
Images, after, 360

Images, successive, of different colours,
the same as superposed colours, 130
Impressions, diagram of visual, 25
Impressions, duration of visual, 24
Incomplete red and green blindness,

293
Intensity of spectrum colours, 74
Intensity of spectrum colours trans-
mitted through chromate of potas-
sium, 78
Interlacing of coloured discs, 133
Iridescent colours, measurement of, 85

Jn.'s determination of colour blind-
ness, 307

Law connecting angular aperture with

extinction, 172
Lens of the eye, 14
Light, luminosity of, coming through

different apertures, 180
Light, the fundamental sensation of

colour, 20
Logarithmic curve of the extinction of

light in spectrum made by an eye

with excess of pigment, 172
Logarithmic curves of extinction of

light, 170
Luminosity, alternative method of

measuring, 93
Luminosity, comparison of, of two

pigments, 87
Luminosity curve of a completely red

blind eye, 286
Luminosity curve of normal and colour

blind vision of equal areas, 381-3
Luminosity curves obtained with

fatigued retina, 384, 386
Luminosity, measurement of, 86
Luminosity of feeble spectrum, 96
Luminosity of pigment by means of

colour discs, 137
Luminosity of pigments in artificial

light, 89
Luminosity of pigments in daylight,

138
Luminosity of spectrum colours on the

fovea, 93
Luminosity of spectrum colours on

yellow spot, 89

INDEX

417

Lnminosity of spectrnm colours outside

yellow spot, 91
Luminosity of spectrum formed by

light of low grade, 102
Luminosity spectrum colour curves, 94

H.'s luminosity curve, 350

Macula lutea, 13

Hatches of *' fatigue" with spectrum

colours, 371, 375, 377
Matching the colour of chromate of

potash by a single ray, 323
Matching the D light by red and green

in the spectrum, 321
Maxwell, 212

Maxwell's colour curves, 227
Maxweirs colour box, 223
Maxweirs colour equations, 225
Maxwell's slit apertures and luminosi-
ties, 228
Measurement of amount of colour

blindness, first method, 296
Measurement of amount of colour

blindness, second method, 802
Measurement of contrast colour seen

in white, 118
Measurement of intensity, alternative

method, 80
Measurement of intensity from pig-
ments, disc method, 82
Mistakes in selection of colours by

colour blind, water-colour washes as

tests, 401
Modification of colour patch apparatus

to measure intensity, 74
Monochromatic images, 49
Monochromatic vision, 338
Monochromatic vision, a probable case

of, 354

N., the luminosity curve of, incom-
pletely green blind, 300

N. W.'s luminosity curve, 345

Nemst lamp spectrum sensation curves
of, 243

Newton's spectrum colours, 272

Normal spectrum colours described by
colour blind person, 273-6

Normal spectrum, sensation curves of,
246

Numerical registration of colour, 126-7

Obstbuotion and absorption factors, 6
Obtaining dominant colour by means
of a single slit, 127

Pabaffin light and pigments, 252-3
Particles, fine, in the atmosphere, 55
Patches of red, green, and blue, effect

of reducing luminosity of, 145
Pendulum curves, 218

Percentage of sensations in different

spectrum colours, 238, 240
Perimeter, 192-4

Permanganate of potash, colour of, 130
Persistency curve from the extinction

curve light (illuminated retina), 188
Photog^phic and mechanical examples.

of effect of nonsynchronous rays, 215
Pigment colours, 6
Pigmentation, excess of, in the yellofr

spot, 349
Pigmentation of yellow spot, lack of,

343
Points of light, when separated, 13
Polarisation of beam passing through

a candle fiame crater of the positive

pole of arc light, 61
Polarisation of scattered light, 60
"Purkinje effect," 146
** Purkinje effect " similar to effect of

light on a photc^^phic film, 221

Rapid fatigue, 395

Ratio of red to green sensation in the
spectrum, 324

Ray, single, to give luminosity of sun-
light, 58

Rayleigh's match to " D " light, 321

Rays for different spectrum intensity,
luminosity of, 100

Rays of nonsynchronous vibration,
effect of, 215

Red and green complete colour blind-
ness, 276

Reflection by a bundle of glasses, 76

Resonator curves, 218

Retina, 11

Rods and cones, 12

Sectob apparatus, 69

Sensation curves and white in terms of

luminosity. 243
Sensation carves, checked by addition

of white, 263
Sensation curves of colour blind similar

to those of normal colour vision, 294
Sensation curves of equal area, 240
Sensation curves of the complete red

and p^reen colour blind, 281
Sensation curves, shifted, 327
Sensations absent in complete colour

blindness, 276
Sensations, colour, 229
Sensations, colour, in colour discs, 248
Sensations, colour, in emerald green,

251
Sensations, colour, in vermilion, 250
Sensations, colour, not identical with

colours, 230
Silver bromide, effect of spectrum on,

217

2d

418 RESEARCHES IN COLOUR VISION

Silver chloride, effect of spectmm on,
216

Simuitaneoos contrast colours, 113

Simultaneous contrast, as a test of
colonr blindness, 410

Simultaneous contrast, measures of,
118-20

Skeins, colours of tests, in wool test,
399

Skj light, 59

Slits for the spectrum, 40, 41

Source of light to use with colour patcl)
apparatus, 53

Spectra, two, apparatus for using simul-
taneously, 42

Spectrum, a white (Miss W.)> 352

Spectrum colours, matched with pig-
ments, 4

Spectrum colours to the normal eye
fatigued and unfatigued, 394

Spectrum, extent of field for different
rays of, 206

Spectrum luminosity to the colour
blind, 279

Spectrum, scalings, 48

Stereoscopic arrangement for viewing
fatigue colours, 365

Stimulations, equal, of the three sensa-
tions, 231-2

Stimulus, equal, percentage curves,
367-9

Sum of separate luminosities equal to
combined luminosity, 105, 106

Sunlight, 54, 58

Surface, receiving, for colour patch, 46

Testing for colour blindness, 398
Tests of colour blindness by simultan-
eous contrast, 410
Tests with dots of colour, 406
Tests with spectrum colours for oolour
blindness, 405

Theory of colour vision, 211

Top, Benham's, 28

Trichromatic theory and Helmholtz,

213
Trichromatic theory and Maxwell,

212
Trichromatic theory and Young, 212

Uncommon, some cases of, colour
blindness, 338

Vebmilion, colour sensations in, 250
Violet fatigue, 391
Vision, colour, theory of, 211
Vision, phenomena in, 21
Vision, recurrent, 21
Vision, zone of distinct, 13
Visual impulses, seat of, 15
Visual receiving apparatus, 220

W., the luminosity curve of (incom-
pletely red blind), 298

Watson and luminosity curves, 381

Watson's flicker wheel, 108

Wave-lengths of Standard Scale
Numbers, 97

White, addition of, changes hue of
colour, 255-6

White, addition of, masked by colour,
264-5

White, percentage of, in green rays,
235

Wool test, 398

X., the luminosity curves of (completely
red blind), 286

Y/s colonr blindness, 314
Young, 212
Young effect, 21

Z.'s luminosity curve, 304

UNIV. OF MICHIGAN,

MAR 2 1 1918

Printed by Ballantyne, Hanson &> Co.
Edinburgh 6r* London